Research On Staring Correlated Imaging Of Block Sparse Target

The azimuth resolution is limited by antenna aperture size in the conventional staring imaging radar.While microwave staring correlated imaging radar has the potential application for high resolution staring imaging in key areas by the interaction of two-dimensional temporal-spatial stochastic radiation field and the radar target.Existing microwave staring correlated sparse recovery technology is mostly absorbed in the problem of coefficients reconstruction of isolated scattering points.While for block sparse target,where strong scattering points are of great quantity and clustered,traditional compressed sensing methods usually reveal poor reconstitution result.In this paper,the structural property of block sparse target itself is utilized to study the correlated imaging algorithms.And the optimum design method of radiant sources for correlated imaging radar is proposed to obtain high resolution recovery image of block sparse target.Firstly,the block sparse imaging technique with combined constraints of independent scattering points and edges is researched.Besides the sparse prior of strong scattering points,the edges' continuity property should also be considered for block sparse target.Thus a combination of negative exponential restraint and total variation(NER-TV)algorithm is proposed in this paper.A sequential order one negative exponential function is introduced to measure the sparsity,while the 2D total variation technique is added to design a novel optimization problem for block sparse target imaging.The novel model can extract structured information of block sparse target.Simulation results show that the proposed method has lower imaging error than the existing algorithms.Secondly,the block sparse imaging technique based on parameter coupling is studied.Considering the structure property of block sparse target,where there exist relationships and dependencies between the nonzero coefficient and its neighboring coefficients,we propose a clustered sparse Bayesian learning with Laplace prior(La-CSBL)algorithm.In the traditional sparse Bayesian learning model,the sparsity of the signal component is controlled by its own corresponding hyperparameter.However,in the new imaging model,hyperparameters still control the sparsity of the signal,while neighboring signal coefficients are correlated with each other through their shared hyperparameters simultaneously.Then the cyclic minimization(CM)method is applied to acquire the solutions of hyperparameters and signal coefficients,and obtain the target image thereby.Simulation results show that the proposed La-CSBL algorithm has lower imaging error than the existing methods.Finally,the optimum design method of radiant sources for correlated imaging radar is studied.In order to acquire high resolution imaging result of block sparse target,it is crucial to ensure the high randomness feature of radiation field under the mechanism of microwave staring correlated imaging.And the formation of stochastic radiation field depends on the design of random radiant sources.In this paper,we put forward the concept of radiation source distribution entropy to describe quantitatively the stochastic characteristics of radiant sources.The parameters optimization of position and frequency of radiant sources can be utilized to maximize the radiation source distribution entropy.Simulation results show that the proposed method can effectively improve the imaging quality.