Multi-Mode Radar Imaging Methods For Maneuvering Platforms

Synthetic aperture radar(SAR)imaging system is a powerful supplement to traditional radar detection,where two-dimensional(2-D)microwave images of illuminated area can be produced and high-resolution details of targets can be extracted.Therefore,SAR image formation has been widely used in multi-mode remote sensing.When SAR imaging system is mounted on maneuvering platforms,reverse constraint on trajectory planning should be alleviated,which makes outstanding contributions to remote sensing.However,existing references on terrain observation by progressive scans(TOPS)SAR and super-resolution forward-looking imaging mostly focus on uniform linear motion platforms,and analysis of maneuvering platforms is still incomplete.Consequently,multi-mode radar imaging methods for maneuvering platforms are researched in this dissertation,where maneuvering TOPS SAR issue and maneuvering forward-looking imaging issue are handled.Main contents of this dissertation include the following four aspects.(1)Spatially variant phase of 3-D acceleration is coupled in maneuvering highly-squinted TOPS SAR model.To solve this problem,a generalized resampling method is proposed.After pre-processing and unfolding operation,regularized 2-D spectrum is reconstructed.Subsequently,generalized resampling mapping eliminates the spatially variant acceleration phase and simultaneously equalizes Doppler parameters of targets within the same range cell.When maneuverability of the platform is high enough,the proposed method significantly outperforms conventional TOPS SAR methods because residual acceleration components in each order Doppler parameter are formulated on the basis of traditional uniform acceleration compensation.(2)Traditional non-linear chirp scaling(NCS)method is capable of eliminating only two terms of spatially variant Doppler parameter.To solve this problem,an approximation-free correction method for first-order spatial variance is proposed.Based on generalized wide-swath SAR processing,the first partial derivative with respect to azimuth spatial variance is derived by envelope-straightened phase.After removing constant term and normalizing,resampling mapping relationship for slow time is constructed,which is capable of eliminating all first-order spatial variance of Doppler parameters at once.The proposed method does not employ any zero-padding,and the correction effect of spatial variance is not restricted by the number of undetermined coefficients.Under some highly maneuverable SAR imaging model,the proposed method performs more robust than traditional NCS method.(3)For monostatic radar system mounted on maneuvering platforms,forward-looking imaging is a difficult issue to be handled.To solve this problem,a super-resolution forward-looking imaging method is proposed.After pre-processing,envelope of targets is concentrated within a single range cell,so that sparse signal reconstruction kernel can be employed to break through the limitation of azimuth real aperture on cross-range resolution.Correspondingly,an over-complete dictionary is optimized and a linear regression model is formulated on the basis of the pre-processing echo.Subsequently,sparsity adaptive matching pursuit(SAMP)kernel is adopted to reconstruct the forward-looking scene,where the halting condition is improved to be more robust and more suitable.The proposed method breaks through application limitations of forward-looking imaging on maneuvering platforms.(4)Aiming at maneuvering forward-looking imaging model,a reasonable super-resolution forward-looking imaging method is proposed relying on sparse Bayesian learning theory.On the basis of pre-processing and the optimized dictionary,explainable prior probability and likelihood probability models are established for forward-looking scene and observed echo,respectively.Subsequently,latent variables in probability model are estimated by a kind of data mining algorithm and simplified by actual physical meaning.Eventually,maximum likelihood estimation of forward-looking scene can be formulated.When dealing with maneuvering monostatic forward-looking imaging issue,the proposed method shows better reconstruction effect under several SNR conditions.