CN104391285A - Self-focusing method of low-frequency ultra wide band SAR (synthetic aperture radar) based on image domain - Google Patents

Abstract

The invention provides a self-focusing method of a low-frequency ultra wide band SAR (synthetic aperture radar) based on an image domain. A technical scheme adopted by the invention is that the self-focusing method comprises the following steps of: firstly, dividing an image of the low-frequency ultra wide band SAR into a plurality of sub images along the azimuth and the distance; then, processing each sub image by adopting a phase error estimating and compensating method based on three-step reconnaissance to obtain high-quality focused sub image; and finally, jointing all focused sub images by adopting a jointing method based on reference point alignment so as to obtain the self-focused image of the low-frequency ultra wide band SAR. The self-focusing method can be used for effectively improving the image quality of the low-frequency ultra wide band SAR.

Description

Based on the low-frequency ultra-wideband SAR auto-focus method of image area

Technical field

The invention belongs to SAR (Synthetic Aperture Radar, synthetic-aperture radar) imaging field, relate to a kind of auto-focus method being applicable to low-frequency ultra-wideband SAR.

Background technology

Low-frequency ultra-wideband SAR technology is one of study hotspot in SAR field.Low-frequency ultra-wideband SAR adopts the work of low-frequency ultra-wideband signal, and have very strong leafage or ground penetrating ability, can detect hidden target, therefore low-frequency ultra-wideband SAR is significant in military and civilian field.

Airborne low-frequency ultra-wideband SAR has large accumulation angle and long synthetic aperture, is vulnerable to the impact of flow perturbation and departs from Desired Track in practical flight process.In order to avoid SAR image defocuses, the precision of sensor needs to reach sub-wavelength level usually, and this is very harsh for most of sensor.Therefore, except utilizing the metrical information of sensor and compensating echo, also need to adopt auto-focus method estimate phase error and compensate, to improve the picture quality of low-frequency ultra-wideband SAR in imaging process.

According to the difference adopting phase error model, auto-focus method is divided into the self-focusing based on echo territory and the auto-focus method based on image area.Auto-focus method based on echo territory from SAR echo, extracts irreducible phase errors complete phase error compensation, but these class methods can only estimate low order phase error, can not estimate order phase error; Auto-focus method based on image area from SAR image, extracts phase error complete phase error compensation, and owing to not relying on phase error model, therefore it can estimate arbitrary order phase error effectively, thus is widely used in SAR imaging field.

The existing auto-focus method based on image area can only meet the self-focusing of high-frequency narrow-band SAR, and is not suitable for low-frequency ultra-wideband SAR situation.Therefore, the self-focusing how solving low-frequency ultra-wideband SAR is a technical matters urgently to be resolved hurrily just.

Summary of the invention

The invention provides a kind of low-frequency ultra-wideband SAR auto-focus method based on image area, to improve the picture quality of low-frequency ultra-wideband SAR.

The basic ideas of technical solution of the present invention are: first, to low-frequency ultra-wideband SAR image along orientation to distance to being divided into several subimages; Then, adopt and based on the phase error estimation and phase error of three step reconnaissances and compensation method, each subimage is processed, obtain high-quality focuson image; Finally, adopt the joining method based on reference point alignment to splice all focuson images, obtain low-frequency ultra-wideband SAR self-focusing image.

The technical solution used in the present invention comprises following treatment step:

The wavelength of known low-frequency ultra-wideband SAR is λ, and antenna beamwidth is θ _a, azimuthal resolution is ρ _a, range resolution is ρ _r, antenna phase center is R to the distance at mapping band center _c.

The first step, subimage divides;

By low-frequency ultra-wideband SAR image along orientation to distance to being divided into several subimages A _k, k=1,2 ..., M.The azimuth width D of anyon image _awith distance width D _rmeet following formula:

$\left\{\begin{array}{c}{D}_a\≤2R_c[\tan ({\mathrm{\θ}}_a/2)-\tan \left(\arcsin (\mathrm{\λ}/4{\mathrm{\ρ}}_a)\right)]\\ D_r\≤512{\mathrm{\ρ}}_r\end{array}\right.$

Overlapping widths L in orientation between subimage _ameet following formula:

L _a≥16ρ _a

Second step, based on subimage phase error estimation and phase error and the compensation of three step reconnaissances;

Following step is carried out to any width subimage:

1. carry out first time to the range unit of subimage to select.

First following formula is utilized to calculate the contrast of subimage range unit:

$B_i=\frac{\sqrt{E[S_i^2\left(j\right)-E[S_i^2\left(j\right)]}}{E{[S}_i^2\left(j\right)]}$

Wherein, B _irepresent the contrast of subimage i-th range unit, represent the intensity of a jth pixel in subimage i-th range unit, E [] is averaging operation symbol.

Calculate the average contrast of subimage:

$\stackrel{\‾}{B}=E\left(B_i\right)$

Then comparative selection degree is greater than the subimage range unit of average contrast, as the subimage range unit that step is 1. selected.

2. carry out second time to the range unit of subimage to select.

First 1. selected to step subimage range unit justifies displacement and windowing process, then utilizes following formula to calculate the mass parameter of subimage range unit center strong scattering point:

$H_m=\frac{1}{D_s}\underset{j=1}{\overset{D_s}{\mathrm{\Σ}}}{S}_m^2(\frac{D-D_s}{2}+j)/\frac{1}{D-D_s}\left(\underset{j=1}{\overset{(D-D_s)/2}{\mathrm{\Σ}}}\right[S_m^2\left(j\right)+S_m^2(D-j+1)\left]\right)$

Wherein, H _mrepresent the mass parameter of m range unit center strong scattering point, D≤128 ρ _arepresent the width of window function, D _s≤ 10 ρ _arepresent the width of strong scattering point protecting energy unit, above-mentioned two parameters are determined according to actual conditions.

Calculate the average quality parameter of subimage:

$\stackrel{\‾}{H}=E\left(H_m\right)$

Mass parameter is finally selected to be greater than the subimage range unit of average quality parameter, as the subimage range unit that step is 2. selected.

3. the subimage range unit comprising edge strong scattering point is rejected.

First in the subimage range unit that step is 2. selected, find energy and divided the strong scattering point blocked by subimage, then reject and comprise the subimage range unit of these strong scattering points, the subimage range unit of not rejecting is as the 3. selected subimage range unit of step.

4. subimage phase error estimation and phase error.

First 3. selected to step subimage range unit carries out orientation to inverse Fourier transform, obtains the value F of a n-th subimage range unit jth pixel in Range compress territory _nj (), then utilizes following formula estimating phase error Θ _ε(j):

${\mathrm{\Θ}}_{\mathrm{\ϵ}}\left(j\right)=\underset{p=1}{\overset{j}{\mathrm{\Σ}}}\underset{n}{\mathrm{\Σ}}\mathrm{Im}\{F_n^{*}\left(j\right)\·{\stackrel{\·}{F}}_n\left(j\right)\}/{\mathrm{\Σ}}_n{\left|F_n\left(j\right)\right|}^2$

Wherein, for F _nthe derivative of (j), for F _nj the conjugation of (), Im{} is for asking imaginary-part operation.

5. subimage phase error compensation and iteration.

First step 4. middle phase error Θ is removed _εj the fixterm of () and linear term, obtain final phase error and then utilize final phase error to carry out the phase error compensation of subimage in Range compress territory, finally carry out fourier transform of azimuth, obtain focuson image.

3rd step, based on the subimage splicing of reference point alignment;

Specific implementation step is:

Step is (i): at original sub image A _ksome the range units (usually getting 10-20 bar) that middle selection energy is larger, k=1,2 ..., M, to each range unit selected according to the data beyond the width D rejecting window function of window function;

Step is (ii): in focuson image, select the subimage range unit (i) corresponding with step, and carry out relevant treatment with the subimage range unit that (i) step is selected, obtain the azimuth deviation amount of every strip image distance unit, then using the azimuth deviation amount of the average of each strip image distance unit azimuth deviation amount as focuson image.

Step is (iii): the azimuth deviation amount according to every width focuson image is spliced focuson image, obtains final low-frequency ultra-wideband SAR self-focusing image.

The invention has the beneficial effects as follows:

(1) improve the precision of phase error estimation and phase error and compensation.When estimator image phase error, three times are carried out to the subimage range unit utilized and has selected, improved the precision of phase error estimation and phase error and compensation.(2) improve the precision of subimage splicing.By selecting the larger range unit focuson image that carries out as a reference point of energy to splice, thus obtain high-quality low-frequency ultra-wideband SAR image.

Accompanying drawing explanation

Fig. 1 is the schematic flow sheet of the low-frequency ultra-wideband SAR auto-focus method that the present invention is based on image area;

Fig. 2 is the low-frequency ultra-wideband SAR image of pending Autofocus processing;

Fig. 3 adopts based on energy maximal criterion phase error estimation and phase error and penalty method and the direct result that obtains of subimage splicing method to Fig. 2;

Fig. 4 is the result adopting the present invention to obtain to Fig. 2.

Embodiment

Below in conjunction with accompanying drawing, the present invention is further explained.

Fig. 1 is the schematic flow sheet of the low-frequency ultra-wideband SAR auto-focus method that the present invention is based on image area.As shown in Figure 1, whole flow process comprises three treatment steps: the first step, and subimage divides; Second step, based on subimage phase error estimation and phase error and the compensation of three step reconnaissances, if high-quality focuson image still can not be obtained through a phase error estimation and phase error with compensation, then need to carry out loop iteration, repeat the iteration of this step until obtain high-quality focuson image; 3rd step, based on the subimage splicing of reference point alignment.

The low-frequency ultra-wideband SAR auto-focus method that the present invention is based on image area is verified by measured data process, and theoretical analysis and measured data result demonstrate validity of the present invention.

Carry out Autofocus processing to certain pattern-band low-frequency ultra-wideband SAR image, input picture size is 827.5m × 1185m (orientation to × distance to), and orientation is 1492 to counting, and distance is 2884 to counting.Image is as the criterion to every 256 points, distance to every 512 along orientation and is divided into 42 subimages, overlapping 16 points in orientation between every two width subimages.

Fig. 2 is the low-frequency ultra-wideband SAR image of pending Autofocus processing.Wherein, horizontal direction be distance to, vertical direction be orientation to.Can be found by Fig. 2, because sensor is accurately not high enough, cause the target in imaging scene to fail to realize focusing on, orientation is serious to defocusing, and the quality of image can not be satisfactory.

Fig. 3 is to the phase error estimation and phase error of Fig. 2 employing based on energy maximal criterion and the result of penalty method and directly subimage splicing method acquisition.Wherein, horizontal direction be distance to, vertical direction be orientation to.Compared with Fig. 2, process based on the phase estimation of energy maximal criterion and error compensating method owing to have employed, in Fig. 3, most of scene objects achieves focusing, but partial subgraph picture is along orientation to being moved, thus causes the stitching error between subimage.In Fig. 3, the scene objects in real rectangle does not focus on, and serious inconsistent phenomenon appears in the railway in imaginary ellipse.

Fig. 4 is the result adopting the present invention to obtain to Fig. 2.Wherein, horizontal direction be distance to, vertical direction be orientation to.Compared with Fig. 3, process based on the phase error estimation and phase error of three step reconnaissances and compensation method owing to adopting, the scene objects in Fig. 4 achieves well focussed substantially.Utilize reference point method of aliging to carry out splicing to subimage, subimage is eliminated substantially along orientation to the phenomenon be moved.In Fig. 4, the scene objects in real rectangle focuses on substantially, and the railway in imaginary ellipse does not occur inconsistent phenomenon substantially, and precision and the smoothness of subimage splicing strengthen greatly.

The above is only the preferred embodiment of the present invention; it should be pointed out that for those skilled in the art, under the premise without departing from the principles of the invention; can also make some improvements and modifications, these improvements and modifications also should be considered as protection scope of the present invention.

Claims (1)

1. based on a low-frequency ultra-wideband SAR auto-focus method for image area, SAR refers to synthetic-aperture radar, and the wavelength of known low-frequency ultra-wideband SAR is λ, and antenna beamwidth is θ _a, azimuthal resolution is ρ _a, range resolution is ρ _r, antenna phase center is R to the distance at mapping band center _c, it is characterized in that, also comprise the steps:

The first step, subimage divides:

By low-frequency ultra-wideband SAR image along orientation to distance to being divided into several subimages A _k, k=1,2 ..., M; The azimuth width D of anyon image _awith distance width D _rmeet following formula:

$\left\{\begin{array}{c}{D}_a\≤2R_c[\tan ({\mathrm{\θ}}_a/2)-\tan \left(\arcsin (\mathrm{\λ}/4{\mathrm{\ρ}}_a)\right)]\\ D_r\≤512{\mathrm{\ρ}}_r\end{array}\right.$

Overlapping widths L in orientation between subimage _ameet following formula:

L _a≥16ρ _a

Second step, subimage phase error estimation and phase error and compensation based on three step reconnaissances:

Following step is carried out to any width subimage:

1. carry out first time to the range unit of subimage to select:

First following formula is utilized to calculate the contrast of subimage range unit:

$B_i=\frac{\sqrt{E[S_i^2\left(j\right)-E[S_i^2\left(j\right)\left]\right]}}{E\left[S_i^2\left(j\right)\right]}$

Wherein, B _irepresent the contrast of subimage i-th range unit, represent the intensity of a jth pixel in subimage i-th range unit, E [] is averaging operation symbol;

Calculate the average contrast of subimage again:

$\stackrel{\‾}{B}=E\left(B_i\right)$

Then comparative selection degree is greater than the subimage range unit of average contrast, as the subimage range unit that step is 1. selected;

2. carry out second time to the range unit of subimage to select:

First 1. selected to step subimage range unit justifies displacement and windowing process, then utilizes following formula to calculate the mass parameter of subimage range unit center strong scattering point:

$H_m=\frac{1}{D_s}\underset{j=1}{\overset{D_s}{\mathrm{\Σ}}}{S}_m^2(\frac{D-D_s}{2}+j)/\frac{1}{D-D_s}\left(\underset{j=1}{\overset{(D-D_s)/2}{\mathrm{\Σ}}}\right[S_m^2\left(j\right)+S_m^2(D-j+1)\left]\right)$

Wherein, H _mrepresent the mass parameter of m range unit center strong scattering point, D≤128 ρ _a, D represents the width of window function, D _s≤ 10 ρ _a, D _srepresent the width of strong scattering point protecting energy unit, above-mentioned two parameters are determined according to actual conditions in the scope limited;

Calculate the average quality parameter of subimage again:

$\stackrel{\‾}{H}=E\left(H_m\right)$

Mass parameter is finally selected to be greater than the subimage range unit of average quality parameter, as the subimage range unit that step is 2. selected;

3. the subimage range unit comprising edge strong scattering point is rejected:

First in the subimage range unit that step is 2. selected, find energy and divided the strong scattering point blocked by subimage, then reject and comprise the subimage range unit of these strong scattering points, the subimage range unit of not rejecting is as the 3. selected subimage range unit of step;

4. subimage phase error estimation and phase error;

First 3. selected to step subimage range unit carries out orientation to inverse Fourier transform, obtains the value F of a n-th subimage range unit jth pixel in Range compress territory _nj (), then utilizes following formula estimating phase error Θ _ε(j):

${\mathrm{\Θ}}_{\mathrm{\ϵ}}\left(j\right)=\underset{p=1}{\overset{j}{\mathrm{\Σ}}}\underset{n}{\mathrm{\Σ}}\mathrm{Im}\{F_n^{*}\left(j\right)\·{\stackrel{.}{F}}_n\left(j\right)\}/{\mathrm{\Σ}}_n{\left|F_n\left(j\right)\right|}^2$

Wherein, for F _nthe derivative of (j), for F _nthe conjugation of (j), Im{} is for asking imaginary-part operation;

5. subimage phase error compensation and iteration;

First step 4. middle phase error Θ is removed _εj the fixterm of () and linear term, obtain final phase error; Then utilize final phase error to carry out the phase error compensation of subimage in Range compress territory, finally carry out fourier transform of azimuth, obtain focuson image;

3rd step, the subimage splicing based on reference point alignment:

Specific implementation step is:

Step is (i): at original sub image A _ksome the range units that middle selection energy is larger, the number of range unit, between 10 to 20, rejects data beyond window function to each range unit selected according to the width D of window function;

Step is (ii): in every width focuson image, select the subimage range unit (i) corresponding with step, and carry out relevant treatment with the subimage range unit that (i) step is selected, obtain the azimuth deviation amount of every strip image distance unit, then using the azimuth deviation amount of the average of each strip image distance unit azimuth deviation amount as focuson image;

Step is (iii): the azimuth deviation amount according to every width focuson image is spliced focuson image, obtains low-frequency ultra-wideband SAR self-focusing image.