Study On High-resolution ISAR Imaging Of Maneuvering Targets In Complex Environments Based On Sparse Bayesian Learning

Inverse Synthetic Aperture Radar(ISAR)can obtain two-dimensional high-resolution images by transmitting wide-band signals and using the relative motion between the target and the radar.High-resolution ISAR images can be used for target feature extraction and recognition.Most of the existing ISAR imaging methods assume that the motion of the target is stable,at this time,RD algorithm can be used to obtain high quality images.But for noncooperative targets,the motion is unknown,and it is likely to be maneuvering.At this time,the imaging quality of the existing methods will rapidly decline or even fail.And in actual situations,ISAR imaging often faces problems such as echo defect,low signal-tonoise(interference)ratio,and target off-grid.The method based on sparse Bayesian learning can make full use of the characteristics of the target and the environment,and provides a new idea for high-resolution ISAR imaging of maneuvering targets.The thesis focuses on the research of high-resolution ISAR imaging of maneuvering targets in complex environments based on sparse Bayesian learning.The specific content includes:echo and probability modeling of maneuvering targets and complex environments;highresolution ISAR imaging of off-grid maneuvering targets based on parameterized dictionaries;ISAR imaging of maneuvering targets under low signal-to-noise(interference)ratio conditions,etc.Related research will provide theoretical and technical support for improving the imaging capabilities of imaging radar under complex conditions.The main content of this thesis includes three parts:The first part studies the ISAR imaging of maneuvering targets based on fast sparse Bayesian learning.Firstly,the sparse observation model is established,and then a probabilistic graph model is constructed by introducing the Gamma-Gaussian prior to the sparse azimuth image of the target.On this basis,the model parameters are estimated by fast sparse Bayesian learning,the rotation parameters are estimated by the Newton method.Finally,the Monte Carlo experiments and measured data imaging results verify the effectiveness of the proposed algorithm.The second part studies the ISAR imaging of off-grid maneuvering targets based on parameterized dictionaries.First,a sparse observation model of the off-grid maneuvering target is established,and then the spike-and-slab prior is introduced to the image and a uniform distribution prior is introduced to the scattering point Doppler.On this basis,the model parameters can be solved based on variational inference and the Newton method in the iterative.Finally,the Monte Carlo experiments and measured data imaging results verify the effectiveness of the proposed algorithm.The third part studies the ISAR imaging of maneuvering targets under the condition of low signal-to-noise(interference)ratio.At this time,the Gaussian mixture model is used to model the environment,and then the Gamma-Gaussian prior is used to model the sparse azimuth image.Furthermore,the model parameters can be solved based on the MAP-EM method and the Newton method in the iteration.Finally,the Monte Carlo experiments and measured data imaging results verify the effectiveness of the proposed algorithm.