Research On Key Techniques Of High-resolution Wide-swath SAR Imaging Processing

Synthetic aperture radar(SAR) is a high resolution remote sensing imaging systems and has received worldwide attention and application,with the advantages of all-whether and full-time capabilities.However,traditional single channel SAR systems suffer from the minimum amtenna area constraint,which makes it a contradiction to simultaneously obtain wide-swath and high-resolution imaging.Wide swath coverage requires a low pulse repetition frequency(PRF) to avoid range ambiguities,while azimuth high resolution requires a high PRF to suppress azimuth ambiguities.Azimuth multichannel SAR systems utilize multiple receivers to add the sampling frequency in azimuth,which are able to overcome this fundamental limitation.This paper focuses on some key technical problems of azimuth multichannel SAR systems in high-resolution and wide-swath(HRWS) mode.The specific topics of this thesis are as follows:(1)Azimuth multichannel SAR signal reconstruction: Traditional renconstruction methods can be divided into the transfer function method and adaptive reconstruction method.Traditional reconstruction methods obtain the reconstruction coefficients by matrix inversion in range-Doppler domain to suppress Doppler ambiguities in each Doppler bins.Doppler-variant reconstruction coefficients can be converted into constant coefficients by a phase compensation method in range-Doppler domain.Then,we just need matrix inversion only once to suppress azimuth ambiguities,which will greatly reduce the computational load.Since the transfer functions and the steering vectors are different from each other for stationary targets and moving targets,we are not able to simultaneously suppress the azimuth ambiguities of both the stationary targets and the moving targets.This paper presents a digital beamforming technique in azimth time domain,which can transform multichannel SAR signals in stripmap mode into unambiguous spotlight SAR signals.The proposed method does not need matrix inversion and can simultaneously suppress azimuth ambiguities of both moving targets and stationary targets.(2)Channel errors calibration of azimuth multichannel SAR: Channel errors include gain errors,phase errors and position errors,which will significantly degrade the performance of azimuth ambiguities suppression.Tradditional subspace calibration methods construct the convariance matrix in each Doppler bin and then decompose the matrix to obtain the signal/noise subspace.Then,channel errors can be estimated from the relationship between the array manifold and subspace.Traditional subspace calibration methods usually estimate phase error via decomposing a Doppler-variant covariance matrix from one Doppler bin,and then average these errors estimated from several Doppler bins to improve the estimation accuracy,which will result in a large computational load.This paper presents a modified subspace calibration method,which can simultaneously utilize all the Doppler bins to estimate channnel errors and greatly increase the number of training samples to improve calibration accuracy.Besides,the proposed method needs to decompose the covariance matrix only once,which will greatly reduce the computational load.Using the strong targets in the SAR image as the calibration sources,this paper also presents two calibration methods based on the strong scatterer and cross-correlation,respectively.The two calibration methods have a high accuracy and robustness when there exist isolated strong targets in the image.(3)Out-of-band azimuth ambiguities suppression of strong targets: Since the practical antenna patterns are not bandlimited,there exist non-bandlimited Doppler spectra for strong targets,which will result in residual azimuth ambiguities.For azimuth multichannel SAR systems in HRWS mode,the Doppler spectra reconstruction coefficients are constructed by a notch to suppress the bandlimited ambiguities.However,due to limitation of the number of receiving channels,the out-of-band Doppler ambiguities usually will not be suppressed and will diffuse into other Dopper spectra,which will not only cause out-of-band ambiguities but also cause extra in-band ambiguities in SAR images.In order to suppress the out-of-band ambiguities of strong targets,this paper constructs two types of notch filters to filter out the out-of-band signals in range-compressed domain and image domain,respectively.The proposed methods can suppress the out-of-band azimuth ambiguities without increasing the number of receiving channels or changing the system PRF.Besides,the proposed methods can effectively keep the resolution of other targets.(4): Moving target detection and radial velccity estimation in HRWS mode: The number of the receiving channels is equal to the maximum number of Doppler ambiguities and the PRF is much less than the Doppler bandwidth in HRWS mode.The SAR image obtained from only one channel is ambiguous,which makes it much harder for moving target detection and radial velocity estimation.This paper transforms the ambiguous multichannel SAR signals into the unambiguous multilook SAR signals based on the DBF technique and then detects the moving targets based on the multilook processing method.In HRWS mode,the moving target may overlap with quite a few clutters.This paper utilizes the maximum likelihood(ML) method to estimate the radial velocity of the moving target.Based on the position characteristic of the clutters for multichannel HRWS SAR,we transform the multi-dimensional peak searching of MLE into one-dimensional peak searching and partition the matrix to reduce the number of matrix inversion,which greatly reduce the computational load.The proposed methods can successfully estimate the radial velocity,even if the total number of the moving target and clutters is equal to the number of the receiving channels.The innovation points of this paper can be summaried as follows:Firstly,this paper proposes three channel errors calibration methods for multichannel SAR systems in HRWS mode.The proposed improved subspace calibration method can improve the performance of the channel errors estimation in low SNRs.The proposed calibration methods based on cross-correlation and strong scattering source have a high calibration accuracy and robustness when there exist isolated strong scattering sources.Secondly,this paper proposes suppression methods of out-of-band ambiguities of strong targets.The proposed methods utilize the directional differentiation beetween the out-of-band signals of the strong targets and other targets to filter out the non-bandlimited signals and protect the components of other targets.Simulations show that the maximal azimuth ambiguity of the strong target can be suppressed from-20 dB to below-40 dB.Real measurements show that the azimuth ambiguity can be futher weakened 10 dB.Thirdly,this paper proposes the moving target detection method and radial velocity estimation method in HRWS mode.The proposed detection method can realize moving target detection without the redundant channel.The proposed radial velocity estimation method of moving target based on maximum likelihood estimation has a high robustness when there exist quite a few clutters or the SNR is low.