High-resolution Imaging Under Non-stationary Moving Platform

Synthetic Aperture Radar(SAR)is capable of working day/night,all-weather and long-range imaging conditions,with low carrier frequency,it can penetrate vegetation and other hidden targets.Because of the many advantages of SAR,not only in the military field has important value but also in the civilian areas.Such as environmental testing,terrain mapping,disaster assessment and many other applications.With the progress of science and technology,a number of new carrier platforms emerged.Such as unmanned aerial vehicles and other small platforms.These platforms are flexible and difficult to be found,the most important thing is casualties can be avoided.They can complete the investigation work close to the enemy area.However,this kind of small-scale platforms are susceptible to unstable air flow,so the motion trajectory often changes drastically.Because of the small platform,it can not carry high-precision inertial navigation system,so in order to achieve the purpose of high-resolution imaging,data based on the motion compensation method appears particularly important,this thesis has carried out research on this kind of problems,the main contents include:1.According to the range and azimuth resolution,the thesis firstly gives an analysis on the principle of high-resolution SAR imaging.The influence of distance migration on imaging is analyzed,and the commonly used Range-Doppler(RD)and Back Projection(BP)imaging algorithms are introduced.The effectiveness of the algorithm is proved by simulation experiments.2.An improved Map-Drift algorithm for UAV SAR imaging is proposed to solve the problem of severe range-variant motion error.Firstly,the initial coarse image data is obtained by two step motion compensation,range cell migration correction and range matching filtering;secondly,de-ramping to remove the second frequency modulation component in the ideal slope.And samples are divided intoa pair of sub-apertures.Select the appropriate distance unit,based on the princple that doppler rate in frequency domain and the distance of the point target are linear,can well estimate the range-variantmotion error;finally,combined with the random sample consensus(RANSAC)method,the estimation of the phase error function of each range unit is completed with eliminating the singular pointat the same time.Then,compensation for residual phase error.The simulation experiment proves,compared with the traditional MD algorithm,the proposed algorithm has the advantages in both precision and robustness.3.We study a high resolution motion error compensation method which is suitable for BP algorithm,including auto-focusing BP algorithm based on average gradient optimal and minimum entropy.Aiming at the problem that auto-focusing algorithm based on the optimization criterion is inefficient,we proposed an auto-focusing BP algorithm based on three control points.Firstly,the inertial navigation system data is used for BP imaging;secondly,based on the relationship of Fourier transform between image domain and phase history domain,the phase history information can be obtained byinverse Fourier transformingthe image of the area where the control point is located;finally,by solving a quadratic programming problem with linear inequality constraints to give the precise location of the APC.Then,we can get a well focused image by using BP algorithm with the estimated APC.The computation efficiency is greatly reduced compared with the autofucus BP uses the method of optimization to estimated APC.