Research On Non-uniform Three-dimentional Configuration SAR Tomography Based On Compressive Sensing

Multi-baseline SAR tomography technology can get the distribution informationon the height of the target, based on the traditional two-dimensional imaging. Thetarget can be reconstructed in the three-dimensional space. It has great potential inmany scenes such as concealed object detection, urban mapping, inversion of thevertical structure of vegetation, therefore it attracted more and more attention.However in practice, Data storage and transmission is very difficult for thehigh-resolution imaging Distributed-SAR system, if the data is three-dimensionalundersampling, the problem can be solved, but the three-dimensional sparse dataalso increase many other signal processing problems. for the reason that, Multi-baseline SAR tomography research under the three-dimensional data undersamplingis very necessary.For various aspects of the problems in the multi-baseline SAR tomography, thispaper proposed some solutions, the main content is as follows:1. Analysis of multi-dimensional baseline chromatography SAR imaging model,analyzed in distance, azimuth, height appeared up under the influence ofnon-uniform sparse data imaging, and gives the entire chromatographic SARsignal processing system.2. Study at a distance-the existence of two-dimensional high-resolution imagingof azimuth nonuniform undersampling case, given the two-dimensional modelbased on compressed sensing imaging, according to this model were establisheddistance matrix remodeling and orientation matrix to achieve reconstructiondimensional reconstruction of the target under the sparse data. In the frequencydomain from bending correction to achieve, due to the orientation of thedown-sampling of the data no longer meet the Nyquist theorem, leading to dataaliasing occurs when calculating the offset distance Doppler domain for thisproblem, based on estimates from the LMS bending correction algorithm, theinterpolation method to complete azimuth estimates based on the minimum meansquare estimation weights ideological factor, an effective solution to the distanceunder the conditions of bending problems.3. Research the2-D high-resolution imaging under the distributed SAR.Comparedto a single distributed SAR imaging, since its use in different spatial locationsobservation received from a different perspective to the target echo produces afrequency shift, the use of small satellites for each echo can be synthesized in awide Doppler bandwidth so that it can be achieved compared to conventional SAR greater distance/azimuth resolution resolution. This paper proposes theuse of the spectrum synthesis method first use interferometric phaseinstantaneous Doppler spectrum estimation, two images Doppler shift. Then thetwo images based on a balanced spectrum superimposed to extend the spectrumof the Doppler signal, the orientation of the target to enhance the resolutionperformance.4. Study the SAR image registration problem. Analysis of the pros and cons ofseveral mainstream image registration method is proposed according to thecharacteristics of each method of secondary SAR image registration methods forimage registration coarse pixel correlation coefficient method firstly, and thenbased on this achieved through maximum spectral method more accuratealignment, which controls the amount of computation without, but also to ensurea high registration accuracy.5. Research the height-dimensional imaging method under the Non-uniformthree-dimensional array configuration. Aiming at high-level imaging sparsearray does not lead to a limited number of baseline imaging have led to theproblem of poor resolution, proposed tracking method based on adaptivesubspace compressed sensing to improve the performance of highly-dimensionalimaging resolution, compared to the orthogonal matching pursuit algorithm, itcan achieve synchronization iteration test of candidate solutions, avoiding theproblem of accumulated error result, improve the quality of imaging.