Study Of High Resolution Imaging For Airborne Wide Area Surveillance Radar

Synthetic aperture radar(SAR)can be used to obtain high resolution imaging scene.However,the synthetic aperture time is usually very large and the beam is always fixed,which limits the fast wide area surveillance ability from multiple aspects and the high revisit ability to the interesting moving targets.In recent years,airborne wide area surveillance radar has attracted more and more attention because it can guarantee wide surveillance coverage as well as to permit a certain revisit rate of moving targets.The airborne radar system transmits signals by steering a narrow antenna from one look direction to another periodically,and performs imaging in each azimuth direction.A wide area surveillance image can be obtained by stitching the sub-images together.In order to guarantee the wide-area surveillance ability from multiple aspects,the antenna beamwidth should scan 3600° space.To get a high revisit rate to targets,the dwell time for each beam or the coherent processing interval(CPI)in one look direction cannot be very long,which is contradictory to long CPI for high azimuth resolution imaging.Therefore,it is a challenge to achieve high azimuth resolution image under high revisit rate.Meanwhile,when the beam bore sight is the same as the flight path(i.e.the antenna illuminates the forward or the backward of the airplane),the Doppler bandwidth is very small even equal to zero.Therefore,blind zone will occur in the forward or the backward direction,and the forward-looking imaging quality degrades greatly.Since the principle of backward-looking imaging is the same as the forward-looking imaging,we refer both of them as forward-looking.Meanwhile,when the sub-images are stitched together,the stitching quality could be severely affected by the non-ideal conditions(i.e.airflow)of the platform,especially for the small unmanned aerial vehicles(UAVs).High resolution imaging methods as well as other key technologies for wide area surveillance imaging radar are considered in this dissertation,and the main contents of the dissertation can be summarized as follows:1.The cross-range resolution is not very high for Doppler Beam Sharpening(DBS)in wide area surveillance radar,and two methods based on spectrum estimation are proposed to solve it.Conventional DBS imaging algorithms images the ground scene based on fast Fourier transform(FFT),and the imaging result is the convolution of real spectrum with the windowed function.When the length of windowed function is limited,the spectrum resolution decreased and the DBS imaging quality degrades.The superresolution model for DBS imaging from the perspective of spectrum estimation is constructed and the up limit for superresolution ability is analyzed.Since spectrum estimation technique can be used to obtain finer spectrum information and low sidelobes,the proposed two methods can enhance the cross-range resolution by a factor of 2 theoretically.The Experimental results on simulated and real data verify the effectiveness of the proposed algorithms.2.Based on above discussion,two super-resolution DBS imaging frameworks(SR-DBS and AESR-DBS)by exploiting the property of time coherency and frequency sparsity of DBS echo are proposed.The spectrum of the DBS echo is caused by the movement of the platform,and the Doppler bandwidth is less than the pulse repetition frequency.Therefore,the DBS image is sparse in frequency domain.The antenna of radar illuminates the ground scene consistently in a scanning mode,and the scatterers can be illustrated by more than one beam,accordingly,there is strong coherence between adjacent pulses.The coherence of echos in azimuth direction and the sparsity of DBS image in frequency domain are fully exploited,and DBS imaging is converted into a problem of sparse recovery with respect to a redundant dictionary.In this framework,the sharpening ratio in our proposed algorithm can be improved by a factor of 2 to 4 theoretically in comparison with the conventional DBS imaging method.Real data experiments demonstrate that the proposed frameworks can lead to a noticeable performance improvement.3.Blind zone always occurs for airborne single-channel forward-looking radar,and an adaptive sparse Bayesian super-resolution imaging algorithm in combined multiple frames data is proposed to improve the azimuth resolution.Firstly,we expand the sparsity assumption of the scattering scene in the single frame data,and the processing space is changed from low dimension of single beam to high dimension of multiple beams when multiple beams data are processed together.When the scatterers are limited,even if the scatterers are not sparse in single beam space,the scatterers are sparse in the multiple beams space.Secondly,the echo of the forward-looking radar in the Gaussian noise is modeled,and the forward-looking imaging is converted into a problem of signal optimization based on Bayesian formalism.In our framework,the parameters are data driven,and conjugate gradient algorithm is used to solve it.Simulated and real data results demonstrate that the proposed algorithm both can increase the resolution of the forward-looking imaging results and suppress the artifacts.4.The azimuth resolution will be greatly decreased due to the small change of the Doppler frequency when the radar works in the forward-looking mode.A synthetic bandwidth azimuth modulation imaging radar(SBAMIR)is proposed to improve the azimuth resolution for the first time.In SBAMIR,the synthetic bandwidth of the received echo is not only determined by the Doppler bandwidth caused by the Doppler history,but also by the modulation bandwidth caused by modulation of transmitted signal.The Doppler bandwidth is very small in conventional forward-looking radar,accordingly the azimuth resolution is very small.However,during the SBAMIR framework,we do not depend on the Doppler bandwidth,but incorporate azimuth modulation to the transmitted signal with a coded sequence in azimuth direction to achieve the high cross-range resolution after matched filtering.Even though the Doppler bandwidth has little contribution to the synthetic bandwidth in ASFLR,the modulation bandwidth can be utilized to increase the synthetic bandwidth.Simulation results are given to verify the effectiveness of the proposed algorithm.5.The stitching problem of sub-images for wide area surveillance radar is studied,and an efficient imaging stitching algorithm based on affine transform is proposed to compensate the motion error.The stitching quality will be greatly affected due to the non-ideal movement of the radar platform when the Doppler center is well estimated and compensated.To solve it,the non-ideal motion of the platform is modeled as the variation of the platform position and the variation of the platform attitude,and the two factors that affect the stitching quality are analyzed mathematically.A correction model for movement error and a novel stitching framework based on affine transformation are proposed,which can correct and stitch the DBS image effectively.In this framework,the translation error caused by the platform position and the rotation error caused by the platform attitude could be compensated simultaneously,the discontinuity and seams could be eliminated when the sub-image is stitched together,which guarantees the real-time wide area imaging system.Therefore,the stitching quality can be improved.Simulated and measured data results demonstrate the effectiveness of the method.