Research On Regularization Method In Radar Real-aperture Super-resolution Imaging

Real-aperture radar can achieve large-scale imaging,and has a wide range of applications in the fields of aircraft autonomous landing,autopilot,topographic mapping,accurate guidance and so on.However,the azimuth resolution of real-aperture imaging is determined by the antenna aperture.High resolution requires large antenna aperture.In practical applications,due to the limitation of platform size,large antenna aperture is not easy to achieve,resulting in low azimuth resolution of real-aperture imaging,which is difficult to meet the requirements of high resolution.Regularization is an effective way to achieve super-resolution imaging,which can obtain resolution improvement beyond Rayleigh limit without changing hardware conditions.However,in order to realize superresolution imaging by regularization method,we must fully consider the distribution characteristics of the target,study the mathematical representation of the target,and establish the objective function for optimization.At the same time,considering the realtime requirement in practical applications,it is also necessary to study the computational efficiency of the algorithm.An effective acceleration strategy is required to accelerate the algorithm and improve the real-time imaging ability of the algorithm.In order to improve the azimuth resolution of real-aperture radar imaging,the echo model of real-aperture radar imaging is analyzed and established.Then,the regularization-based super-resolution method is studied,and the corresponding acceleration algorithm is proposed to better meet the needs of practical application.The main innovations in the research are as follows:1.A regularization method based on sparse constraint is proposed.By introducing the sparse constraint of targets under the regularization framework and solving the optimization problem,the resolution of adjacent targets in azimuth beam is improved significantly,and the problem of limited resolution improvement by traditional methods is solved.In order to meet the real-time requirements in practical applications,a fast sparse super-resolution method based on second-order vector extrapolation is proposed,which improves the computational efficiency by more than 8 times using vector extrapolation,thus solving the problem of slow convergence of the original algorithm.2.A regularization super-resolution method based on one-dimensional total variation constraint is presented.By introducing the total variation constraint of target under the regularization framework and solving it optimally,the target contour information can be effectively maintained while improving the resolution,which solves the problem of target contour loss in the implementation of super-resolution imaging by traditional methods.3.To further improve the ability of distinguishing adjacent targets by the total variation method,a combination constraint regularization super-resolution method is proposed.Through the complementary advantages of the total variation and sparse constraints,the insufficiency of the ability to distinguish adjacent targets by a single total variation constraint is solved,and the azimuth resolution is further improved while the contour characteristics are maintained to achieve super-resolution imaging.4.In order to solve the problem that the total variation method and the combination constraint method have too high computational complexity in a single iteration due to the inversion of the matrix,a fast total variation regularization method and a fast combination constraint regularization method are presented.The fast inversion of the matrix is achieved by Gohberg-Semencul(GS)representation,which reduces the computational complexity of the total variation method and the combination constraint method by one order.The above methods are verified by simulation and measured data.The proposed method can effectively improve the azimuth resolution of radar real-aperture imaging.