Research On Cross-range Scaling Approaches For Inverse Synthetic Aperture Radar Imaging

Inverse synthetic aperture radar(ISAR),an important high-resolution imaging tool,is widely used in military and civil areas.As the development of target auto-recognition technique,the ISAR images are usually used to extract the target information as the input of recognition systems,which require the ISAR images to provide not only clear shape but also accurate size information about the target.Since the targets of ISAR are always noncooperative and uncontrollable,and the azimuth resolution of ISAR image is derived from the Doppler difference generated by the target's motion,its real azimuth size cannot be acquired until a special process,named cross-range scaling,is completed.In general,the cross-range scaling approach can be divided into two main processes.One is estimation for the motion parameters of target from the echoes or images.The other is to calculate the real size of an azimuth bin,which is called as the cross-range scaling factor,based on the estimated motion parameters and specific imaging theory,then adjust the proportion and mark the real size of the ISAR image.This paper researches the cross-range scaling approaches for the ISAR images of smoothly moving targets and maneuvering targets,respectively.As for the first kind of targets,since the imaging theory of Range-Doppler(RD)algorithm,which is the most common imaging algorithm for them,is direct and simple,the calculation of cross-range scaling factor needn't be paid too much attention.Therefore,the key process of cross-range scaling for these targets is to estimate their motion parameters,i.e.the effective rotation velocities.Considering that the existing estimation methods have limitations on precision,efficiency and adaptability,this paper explores the more accurate estimation methods for rotation velocity,especially for the slowly rotating targets and complicated-structure targets.The slowly rotating targets are usually rotating with small velocity,which makes the existing methods that are limited by the resolution of Fourier transform aren't able to meet the required precision for scaling.Thus,combining the iterative adaptive approach(IAA)with the second order discrete polynomial-phase transform(DPT),this paper proposes a novel estimation method named as IAA-DPT,which is able to accurately estimate the rotation velocity of slowly rotating targets.As for the second kind of targets,it's hard to find isolated scatterers on them when estimating their rotation velocity,so many nonlinear methods always generate estimation errors due to the existence of cross terms.By importing the multi-delay DPT theory(MDPT)and Keystone transform,this paper proposes the MDPT-KT estimation method,which is able to deal with the complicated-structure targets.The two proposed methods are verified as effective and accurate by the experiments on simulated and realmeasured data.With regard to the maneuvering targets,since the imaging theories of the corresponding algorithms are different from RD algorithm,both of the two processes of scaling need to be considered.Firstly,the uniformly accelerating model is chosen to be the effective imaging model of the maneuvering target.After that,this paper builds the echo signal model and analyzes the relationship between the rotation parameters and the phase coefficients of slow-time signals.Based on the phase coefficients estimation and scatterers pairing approach,a novel estimation method for the rotation parameters of maneuvering targets is proposed.Then,aiming to the most common imaging approach for these targets,i.e.the Range-Instantaneous-Doppler(RID)algorithm,this paper derives the calculation for the cross-range scaling factor of the instantaneous RID images,and compares it with that of the RD algorithm.Following that,the imaging and scaling theories of a novel imaging algorithm for maneuvering targets,named as Range-ChirpRate(RCR)algorithm,is discussed in this paper.Combined the cross-range scaling process for RCR images,this paper proposes the solutions for the two main problems existing in RCR imaging theory,i.e.the compensation for image distortion along range dimension and selection for the azimuth adjustment factor,to make up the limitations of the origin RCR algorithm.At last,by simulation and real-data experiments on the maneuvering targets,the aforementioned estimation method,imaging and scaling approaches are verified as effective.