1. A high-resolution spaceborne SAR efficient time-frequency hybrid imaging method is characterized by comprising the following steps:

step one, performing range compression and up-sampling on SAR echo data;

step two, carrying out azimuth sub-aperture division on the SAR echo data processed in the step one, carrying out azimuth back projection on the data in the sub-apertures at the moment, wherein the image obtained in each sub-aperture is a coarse image, N coarse images are obtained in total, and N is the number of the sub-apertures;

performing azimuth Fourier transform on each coarse image to obtain each sub-aperture frequency domain image;

fourthly, overlapping the N sub-aperture frequency domain images to obtain a high-resolution image of a distance time domain-orientation frequency domain;

and fifthly, performing azimuth inverse Fourier transform on the high-resolution image to obtain a full-resolution image.

2. The method of claim 1, wherein the range-wise compressing and sampling the SAR echo data is performed by 4 times up-sampling with frequency-domain zero padding.

3. The method of claim 1, wherein the total sub-aperture bandwidth is higher than the bandwidth B corresponding to the beam itself_{a_β}。

4. The method according to claim 1, wherein the fourth step is specifically:

and constructing a null image meeting the full-aperture sampling interval, and copying the frequency spectrum corresponding to each sub-aperture frequency domain image to the null image to obtain a high-resolution image of a distance time domain-azimuth frequency domain.

5. A high-resolution spaceborne SAR high-efficiency time-frequency hybrid imaging system is characterized by comprising a distance direction compression module, an up-sampling module, a sub-aperture division module, a Fourier transform module, a frequency spectrum superposition module and an inverse Fourier transform module;

the range direction compression module is used for performing range direction compression on the SAR echo data, and the SAR echo data subjected to range direction compression are input to the up-sampling module;

the up-sampling module is used for performing up-sampling operation on the SAR echo data after the distance direction compression, and the SAR echo data after up-sampling is input to the sub-aperture division module;

the sub-aperture division module is used for carrying out azimuth sub-aperture division on the SAR echo data after the up-sampling, at the moment, azimuth back projection is carried out on the data in the sub-apertures, the image obtained in each sub-aperture is a coarse image, N coarse images are obtained in total, and N is the number of the sub-apertures; inputting the N coarse images into a Fourier transform module;

the Fourier transform module is used for performing azimuth Fourier transform on the N coarse images to correspondingly obtain N sub-aperture frequency domain images;

the frequency spectrum superposition module is used for constructing a null image meeting the full-aperture sampling interval, and copying the frequency spectrum corresponding to each sub-aperture frequency domain image onto the null image to obtain a high-resolution image of a distance time domain-azimuth frequency domain;

and the inverse Fourier transform module is used for performing azimuth inverse Fourier transform on the high-resolution image to obtain a full-resolution image which is the final output of the system.

Background

Synthetic Aperture Radar (SAR) is a full-time and all-weather high-resolution microwave remote sensing imaging radar, and can be installed on flight platforms such as airplanes and satellites. The method has unique advantages in the aspects of environmental monitoring, marine observation, resource exploration, crop estimation, mapping, military affairs and the like, and can play a role which is difficult to play by other remote sensing means.

Because the bunching and sliding bunching mode can improve the azimuth resolution by improving the azimuth synthetic aperture time, a clearer SAR image is obtained, high-resolution imaging is an important research direction of the SAR imaging at present, and the research of a high-efficiency imaging method is the current research enthusiasm. Compared with the traditional SAR imaging, the high-resolution imaging is faced with a higher-order slant range model, larger parameter space-variant and more complex system design, and is particularly represented in the following aspects:

the existing imaging algorithms based on frequency domains CS, NCS, NLCS and the like have high efficiency, but when the imaging algorithms are used for processing large-scene high-resolution (close to 0.1m) SAR imaging, high-order azimuth phase errors and space-variant errors cannot be completely corrected, and the image quality is obviously reduced and even defocuses due to the residuals of the errors; in addition, in high-resolution wide swath sliding bunching imaging, a variable repetition frequency system is mostly adopted, so that azimuth direction non-uniform sampling is caused, direct conversion to a frequency domain cannot be realized, interpolation operation cannot be avoided, and errors can be introduced; in addition, the linear change of the rotation of the azimuth beam along with time can not be ensured in the high-resolution wide swath sliding beamforming imaging, which directly results in that the Deramping algorithm can not be applied, i.e. the rotation can not be converted into the azimuth frequency domain, and great difficulty is caused to the application of the frequency domain imaging algorithm.

The existing time-domain-based imaging algorithm is paid attention again because the imaging does not depend on a flight path model and has high imaging precision. Whether the azimuth direction is uniformly sampled or not has no influence on a time domain imaging algorithm, the algorithm does not need to convert data to a frequency domain, and the algorithm bypasses the limitations of variable repetition frequency and incapability of applying a Deramping algorithm. But the BP algorithm operand is O (n)³) The magnitude and huge calculation amount limit the application of the algorithm, and the high-efficiency algorithm needs to be researched. Existing high-efficiency imaging algorithms, such as the FFBP algorithm proposed in l.m.h.ulander 2003, are based onThe BP integral is split, and the calculation amount is greatly reduced by step-by-step synthesis. However, the step-by-step iterative synthesis may cause accumulation of errors to some extent, which affects the imaging quality, and the accumulation is more serious as the number of grading times is more. In addition, the conventional time domain fast imaging algorithm, such as the FBP proposed in Yegulalp a.f.1999, mostly involves interpolation operations in sub-aperture image fusion, coordinate system conversion and image fusion, and it is known that fast interpolation operations have errors, and accurate interpolation operations significantly increase the amount of operations, which deviates from the requirement of efficient imaging.

Therefore, the defect that errors are introduced due to interpolation operation is inevitable for improving efficiency aiming at the efficient BP algorithm.

At present, no scheme for avoiding aperture fusion errors without difference operation exists.

Disclosure of Invention

In view of the above, the invention provides a high-efficiency time-frequency hybrid imaging method and system for a high-resolution spaceborne SAR, which can perform fusion of sub-aperture images through a concept of sub-aperture frequency superposition, and avoid interpolation operation through superposition in a frequency domain in sub-aperture fusion, thereby fundamentally avoiding aperture fusion errors.

In order to achieve the purpose, the technical scheme of the invention comprises the following steps:

the method comprises the steps of firstly, performing range direction compression and up-sampling on SAR echo data.

And step two, carrying out azimuth sub-aperture division on the SAR echo data processed in the step one, carrying out azimuth back projection on the data in the sub-apertures at the moment, wherein the image obtained in each sub-aperture is a coarse image, and obtaining N coarse images in total, wherein N is the number of the sub-apertures.

And thirdly, performing azimuth Fourier transform on each coarse image to obtain each sub-aperture frequency domain image.

And fourthly, overlapping the N sub-aperture frequency domain images to obtain a high-resolution image of a distance time domain-azimuth frequency domain.

And fifthly, performing azimuth inverse Fourier transform on the high-resolution image to obtain a full-resolution image.

Further, the echo data of the SAR is subjected to range direction compression and sampling, wherein the up-sampling is realized by adopting frequency domain zero padding, and is subjected to 4 times of up-sampling.

Further, the total sub-aperture bandwidth is higher than the bandwidth B corresponding to the beam itself_{a_β}。

Further, the fourth step is specifically:

and constructing a null image meeting the full-aperture sampling interval, and copying the frequency spectrum corresponding to each sub-aperture frequency domain image to the null image to obtain a high-resolution image of a distance time domain-azimuth frequency domain.

The embodiment of the invention also provides a high-resolution space-borne SAR high-efficiency time-frequency hybrid imaging system which comprises a distance direction compression module, an up-sampling module, a sub-aperture division module, a Fourier transform module, a frequency spectrum superposition module and an inverse Fourier transform module.

And the range direction compression module is used for performing range direction compression on the SAR echo data, and the SAR echo data subjected to range direction compression is input to the up-sampling module.

And the up-sampling module is used for performing up-sampling operation on the SAR echo data after the distance compression, and inputting the SAR echo data after the up-sampling into the sub-aperture division module.

The sub-aperture division module is used for carrying out azimuth sub-aperture division on the SAR echo data after up-sampling, at the moment, azimuth back projection is carried out on the data in the sub-apertures, the image obtained in each sub-aperture is a coarse image, N coarse images are obtained in total, and N is the number of the sub-apertures; the N coarse images are input to a fourier transform module.

And the Fourier transform module is used for performing azimuth Fourier transform on the N coarse images to correspondingly obtain N sub-aperture frequency domain images.

And the frequency spectrum superposition module is used for constructing a null image meeting the full-aperture sampling interval, and copying the frequency spectrum corresponding to each sub-aperture frequency domain image to the null image to obtain a high-resolution image of a distance time domain-azimuth frequency domain.

And the inverse Fourier transform module is used for performing azimuth inverse Fourier transform on the high-resolution image to obtain a full-resolution image which is the final output of the system.

Has the advantages that:

the invention provides the concept of sub-aperture frequency superposition to fuse sub-aperture images, and the sub-aperture fusion avoids interpolation operation by superposing in a frequency domain, thereby fundamentally avoiding aperture fusion errors. And the algorithm realizes high-efficiency processing by splitting the sub-aperture, and is a high-efficiency algorithm suitable for high-resolution satellite-borne SAR precise imaging.

The imaging method provided by the invention firstly compresses and upsamples the distance direction, the upsampling aims to reduce the error during the operation of the BP in the direction, and the upsampling is realized by adopting frequency domain zero filling. Typically 4 times up-sampling is performed to ensure accuracy. And then, azimuth subaperture division is carried out, and the size and the number of the subaperture division should comprehensively consider the efficiency and the imaging precision. At this time, the data in the sub-apertures are subjected to azimuth back projection, and the obtained image in each sub-aperture is a coarse image because of the lower azimuth resolution. N coarse images are obtained through the operation, and N is the number of the sub apertures. And then, performing azimuth Fourier transform on each image to obtain a distance time domain-azimuth frequency domain image, wherein the frequency spectrums of the sub-apertures are staggered in the azimuth frequency domain due to the fact that the sub-apertures correspond to different squint angles. At the moment, the operation of obtaining the high-resolution image from the low-resolution image fusion is completed by accumulating in the azimuth frequency domain, and the operation has no interpolation operation. And after the N sub-aperture frequency domain images are superposed, obtaining a high-resolution image of a distance time domain-azimuth frequency domain, and performing azimuth inverse Fourier transform to obtain a full-resolution image.

Compared with the prior art, the method can ensure high-efficiency imaging without interpolation operation and has the effect of obtaining more accurate SAR imaging quality.

Drawings

FIG. 1 is a flow chart of a sub-aperture frequency domain based superposition FBP algorithm;

FIG. 2 is a spaceborne SAR beamforming pattern geometry;

FIG. 30.1 m resolution point target location map;

FIG. 4 is a two-dimensional contour map of the target imaging of the inner point of the sub-aperture; wherein (a) the target azimuth spectrogram of a point near the center of the scene; (b) the point target sub-aperture imaging two-dimensional result graph;

FIG. 5 is a two-dimensional contour map of a dot matrix target image. Where (a) (b) (c) (D) are the results of the imaging of the object at points A-D in the scene.

Detailed Description

The invention is described in detail below by way of example with reference to the accompanying drawings.

The invention provides a high-resolution spaceborne SAR efficient time-frequency hybrid imaging method, which comprises the following basic implementation processes:

the method comprises the steps of compressing and upsampling in the distance direction, wherein the upsampling aims to reduce errors in the operation of the BP in the azimuth direction, and the upsampling is realized by adopting frequency domain zero padding. Typically 4 times up-sampling is performed to ensure accuracy.

And then, azimuth subaperture division is carried out, and the size and the number of the subaperture division should comprehensively consider the efficiency and the imaging precision. In this case, the data in the sub-aperture is subjected to the azimuth back projection, and it should be noted that the operation is performed in a three-dimensional rectangular coordinate system. The resulting image within each sub-aperture is a coarse image because of its lower azimuthal resolution. N coarse images are obtained through the operation, and N is the number of the sub apertures. And then, performing azimuth Fourier transform on each image to obtain a distance time domain-azimuth frequency domain image, wherein the frequency spectrums of the sub-apertures are staggered in the azimuth frequency domain due to the fact that the sub-apertures correspond to different squint angles. At the moment, the operation of obtaining the high-resolution image from the low-resolution image fusion is completed by accumulating in the azimuth frequency domain, and the operation has no interpolation operation.

And after the N sub-aperture frequency domain images are superposed, obtaining a high-resolution image of a distance time domain-azimuth frequency domain, and performing azimuth inverse Fourier transform to obtain a full-resolution image.

Fig. 1 is a flow chart of a frequency domain superposition FBP algorithm based on sub-apertures.

Firstly, compressing and upsampling the distance direction by an algorithm, wherein the upsampling aims to reduce the error during the operation of the BP in the direction, and the upsampling is realized by adopting frequency domain zero filling. Typically 4 times up-sampling is performed to ensure accuracy. This is no different from conventional BP imaging algorithms.

And step two, dividing the azimuth subaperture. For the convenience of analysis, as shown in fig. 2, here the imaging geometry is based on the beamforming mode, β is the beam width, θ is the azimuth angle, the satellite always irradiates the same region on the ground during working time, and the sliding beamforming mode can obtain the imaging of several beamforming modes through reasonable splitting. It is clear that the minimum sampling pitch required for imaging is now

Wherein Δ x is the azimuthal sampling interval; Δ y is the distance-wise sampling interval; v. of_g＝v_r ²/V_sIs the wave foot velocity, B is the distance bandwidth; c is the speed of light; b is_{a_total}＝B_{a_β}+B_{a_rot}Bandwidth B introduced by beam rotation for total azimuth bandwidth_{a_rot}Bandwidth B corresponding to the beam itself_{a_β}It is generally recommended that the total sub-aperture bandwidth be slightly higher than the bandwidth B corresponding to the beam itself_{a_β}Twice, the efficiency can be guaranteed to be higher at this moment.

And step three, carrying out sub-aperture division to obtain coarse resolution images, and carrying out the same imaging on each sub-aperture to obtain N coarse resolution images. And performing azimuth FFT on each sub-aperture to obtain a frequency domain map of each sub-aperture image.

And step four, constructing a hollow image meeting the full-aperture sampling interval, wherein it needs to be explained that the image exists in an azimuth frequency domain, and the frequency spectrum corresponding to each sub-aperture image is copied to the hollow image, which is equivalent to finishing the interpolation of the sub-aperture image. Essentially, this interpolation is achieved by frequency domain zero padding. The grid spacing here is different from the sub-aperture image grid spacing, and whether the grid spacing is appropriate depends entirely on whether the current spectrum is aliased or not. Considering that there are spectra of the same frequency domain components between sub-aperture images, the spectra of the frequency domain components are superimposed in the frequency domain rather than overlaid for different sub-aperture images. In the subsequent embodiment, the frequency domain superposition operation is very accurate, the finally obtained full-resolution image is very ideal in the frequency domain spectrum, and indexes of a corresponding target such as time domain resolution, sidelobe ratio and the like are also close to theoretical values.

And fifthly, performing azimuth IFFT on the high-resolution image in the azimuth frequency domain to obtain a full-resolution image. Therefore, high-resolution SAR imaging FBP imaging is realized.

It can be seen that the algorithm has no interpolation operation at all, and does not introduce extra errors and operation amount, the errors of the algorithm only come from operation, and the errors introduced by the up-sampling operation such as frequency domain zero padding are negligible, so that the algorithm can realize accurate high-resolution SAR imaging.

The embodiment of the invention also provides a high-resolution space-borne SAR high-efficiency time-frequency hybrid imaging system which comprises a distance direction compression module, an up-sampling module, a sub-aperture division module, a Fourier transform module, a frequency spectrum superposition module and an inverse Fourier transform module.

And the range direction compression module is used for performing range direction compression on the SAR echo data, and the SAR echo data subjected to range direction compression is input to the up-sampling module.

And the up-sampling module is used for performing up-sampling operation on the SAR echo data after the distance compression, and inputting the SAR echo data after the up-sampling into the sub-aperture division module.

The sub-aperture division module is used for carrying out azimuth sub-aperture division on the SAR echo data after up-sampling, at the moment, azimuth back projection is carried out on the data in the sub-apertures, the image obtained in each sub-aperture is a coarse image, N coarse images are obtained in total, and N is the number of the sub-apertures; the N coarse images are input to a fourier transform module.

And the Fourier transform module is used for performing azimuth Fourier transform on the N coarse images to correspondingly obtain N sub-aperture frequency domain images.

And the frequency spectrum superposition module is used for constructing a null image meeting the full-aperture sampling interval, and copying the frequency spectrum corresponding to each sub-aperture frequency domain image to the null image to obtain a high-resolution image of a distance time domain-azimuth frequency domain.

And the inverse Fourier transform module is used for performing azimuth inverse Fourier transform on the high-resolution image to obtain a full-resolution image which is the final output of the system.

Examples

The simulation realizes the sliding bunching mode lattice imaging with high resolution in the azimuth direction, wherein the scene width A multiplied by R is 2km multiplied by 10 km. The system parameters are shown in table 1.

Table 1. system parameters.

Parameter(s)	Numerical value
		Carrier frequency	X wave band
PRF	The repetition frequency is changed to 2000 Hz-2600 Hz
		β	0.21°
Distance to bandwidth	3000MHz
		θ	11.6°
Working time	39s
		Initial sampling time	Change with migration

Because the mode has very high resolution and a large scene, a system adopts a segmented variable PRF and sampling moment system when designing the mode. This ensures the system design of the high-branch wide-amplitude mode and greatly reduces the effect of migration. The sliding bunch simulation dot matrix data is imaged, and the dot matrix target setting is shown in fig. 3. This scene is approximately a positive side mode illumination with an azimuth angle of 11.6 deg. And (3) imaging by adopting an FBP algorithm based on sub-aperture frequency superposition. Looking at the sub-aperture imaging results, as shown, the azimuthal spectrum is found to be not a rectangular envelope. The scene data is imaged in 45 sub-apertures, each accumulating 0.86 s. The sub-aperture 23 at the center is imaged and transformed to the azimuth frequency domain, and the point target spectrum is shown in fig. 4. At this time, the doppler bandwidth of the target in the target sub-aperture is about 3450Hz, which is substantially equal to the target bandwidth in fig. 4(a), and fig. 4(b) shows the imaging result in the target sub-aperture, which shows that the target focusing effect at this point is good.

And finishing the frequency domain superposition operation and IFFT of the images in each sub-aperture to obtain a full-resolution image and evaluating an imaging result, wherein the evaluation result of the lattice target in the 2 multiplied by 10km (A multiplied by R) width is shown in the table 2, the imaging adopts a-25 dB Taylor window to carry out sidelobe suppression, and finally all targets PSLR and ISLR in the scene meet the imaging requirement. Figure 5 illustrates an imaged two-dimensional map of the targets at various points. Therefore, the target distance and the azimuth direction imaging of each point are ideal, and the imaging quality is not lost because the algorithm does not involve interpolation operation.

And 2, imaging lattice target evaluation results of the sub-aperture frequency domain superposition FBP algorithm.

The above description is only a preferred embodiment of the present invention, and is not intended to limit the scope of the present invention. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.