Study On High-precision ISAR Cross-range Scaling Methods For Moving Targets

Inverse synthetic aperture radar(ISAR)can generate 2-D high resolution images of non-cooperative targets and it has been applied to various fields.Two of the main problems in ISAR imaging are the cancellation of space-variant phase error arising from the rotation and the cross-range scaling.In this thesis,the algorithms to remove rotational space-variant phase error and to achieve cross-range scaling which are suitable for implementation are studied.The contents in this thesis are summarized as follow.(1)The imaging and cross-range scaling algorithm for targets with uniform rotationThe non-parametric motion compensation algorithms bring about the inconsistency between the rotation centers in range and azimuth direction.This problem adversely affects the estimation of the rotation speed and the compensation of rotational space-variant phase error,which will impair the ISAR image quality and the cross-range scaling.In this thesis,after the analysis of the influence of range-variant phase error and the inconsistency between the rotation centers in range and azimuth direction,an ISAR imaging and scaling algorithm based on Local Average Doppler Trend(LADT)is proposed.This algorithm extracts the rotation information of the target from the signals' phase terms.Considering that a single scatter's phase function cannot accurately reflect the rotation influence,we utilize the phase functions of all the scatters in a range cell to form a LADT signal.This signal is able to reflect the Doppler centers in short slow-time intervals caused by rotation.Based on the LADT signals in range cells,the Doppler rates in these range cells can be estimated.Then,according to the relation between Doppler rates and the ranges,the rotation speed of the target is estimated.During the procedure,in order to get rid of the influence of noise and the discontinuity of the extracted phase function,the Random Sample Consensus(RANSAC)method is adopted.After the estimation of the rotation speed,the motion compensation function for range-variant phase error and cross-range scaling are achieved.The proposed method is not affected by the inconsistence between the rotation centers in the range and azimuth direction.The effectiveness and efficiency are demonstrated by the processing of simulated and acquired data.(2)The imaging and scaling algorithm for a maneuvering targetThe rotation of a maneuvering target can be modeled by a rotation with acceleration,which can bring about space-variant phase errors.In the thesis,the Resampling-Nonlinear Least Squares(RS-NLS)algorithm is proposed to estimate the rotation information and to eliminate the space-variant phase error with the estimated rotation information.With accurate rotation information,the space-variant phase error can be removed by the range-variant phase error phase compensation function and resampling.After the removal of the space-variant phase error,the image quality will be optimized.Therefore,this algorithm seeks for the rotation information by solving a nonlinear least squares problem whose objective function is the image quality.This nonlinear least squares problem is solved by Gauss-Newton algorithm.The Gauss-Newton algorithm gradually generates estimates of rotation information which are closer to the real one.In each iteration,the calculations comprise FFT and resampling.Compared with the exited algorithms,the proposed algorithm is more efficient.The effectiveness of the proposed algorithm is demonstrated by the processing of the simulated and acquired data.