Study On High-Resolution Fusion Imaging Method Of ISAR

With the increasing demand for high resolution,multi-angle detection and imaging for space/air targets,high resolution inverse synthetic aperture radar(ISAR)is gradually developing towards bi/multi-static network imaging,multi-polarization and multi-channel imaging,etc.Although multi-perspective and multi-polarization data fusion can improve image resolution and obtain richer target features,they increase the difficulty in ISAR signal processing.At present,how to effectively integrate the information from multi-perspective and multi-polarization data to improve the imaging quality has become a critical problem in the field of ISAR and has received intensive attentions.In order to solve the above-mentioned problems,this thesis firstly establishes the geometric model and signal model for multi-static ISAR,and then studies the image fusion method under coherent and non-coherent conditions.Specifically,it discusses target rotation parameter and observation angle difference estimation,and proposes high-resolution ISAR image synthesis and image enhancement methods.As for full-polarization ISAR imaging,an effective polarization fusion method is studied so as to make full use of the polarization information among channels,and improve the reliability and comprehensiveness of feature extraction.The related research will provide theoretical and technical supports for the improvement of space target detection capabilities of ground-based radars in China.The main content of this thesis can be summarized as follows:In the first part,the geometric model and signal model of multi-static ISAR for space target observation are established,and multi-static ISAR fusion imaging conditions are analyzed.On this basis,multi-static ISAR coherent fusion imaging method in two cases is studied,and the imaging resolution is analyzed: 1)For overlapped radar observation areas,the rotation angle is estimated based on the Doppler difference between two sub-apertures,and the observation angle difference is estimated by scatterers association,then the signal is synthesized coherently to achieve fusion imaging;2)For non-overlapped observation areas,a multi-static ISAR fusion imaging method is proposed based on Bayesian learning in low signal-to-noise ratio.This method transforms the multi-static ISAR fusion imaging problem into a sparse signal reconstruction problem,and then proposes an iterative optimization method.Specifically,the gradient descent is applied to estimate the target rotation angle and the observation angle difference,and the variational inference(VI)is applied to solve the fused image.Finally,the validity of algorithm is verified by simulated data.In the second part,the method of multi-static ISAR high resolution fusion imaging based on non-coherent observation is studied.Firstly,the mapping between ISAR images obtained by different radars is analyzed.Then,sparse observation model is constructed.Finally,a multi-static ISAR fusion imaging method based on improved orthogonal matching pursuit(OMP)is proposed.This method utilizes gradient descent and line search to obtain accurate estimates of the target rotation angle and the angle difference between two radars,and adopts the improved OMP to obtain multi-set ISAR image fusion.Finally,the effectiveness of the algorithm is verified by simulated and measured data.In the third part,high resolution fusion imaging is studied for full polarization ISAR.Firstly,a full polarization measurement matrix and imaging model are constructed according to the consistent property between the signal support region and the sparse basis among different polarization channels.Then,a method based on multi-channel sparse learning via iterative minimization(MC-SLIM)is proposed for high-resolution fusion imaging.This method combines multi-polarized channel signals to solve the sparse imaging problem,i.e.it sparsely reconstructs the ISAR image of each channel and minimizes the energy of a single scatterer between channels,thus can ensure the consistency of target scatterers among polarization channels.Finally,the effectiveness of the algorithm is verified by point simulated data,electromagnetic simulated data,and measured data processing.