UAV SAR Moving Target Detection

Unmanned aerial vehicle synthetic aperture radar(UAV-SAR)is a high-resolution radar based on UAV flight platform.With the development and application of UAV technology,the advantages of small size and easy manipulation make it possible for multi UAV cooperation.However,the multi-platform UAV-SAR has many problems,such as the geometric relationship between platforms is not fixed,and the effective sampling time is short.As a result,the traditional ground moving target identification(GMTI)method is not suitable for UAV-SAR.Based on the above background;this thesis studies the methods of moving target detection and parameter estimation of UAV-SAR.The main work of this thesis includes:1.Aiming at the problem of clutter suppression of multiple UAV,a tensor robust principal component analysis based on total variation is proposed(TV-TRPCA)in this thesis.In this method,multiple data matrices are stacked into three-dimensional tensors,and the moving target echo is separated by using the mathematical properties of each component in the tensor,to achieve the effect of clutter suppression.TV-TRPCA algorithm uses tensor structure to overcome the disadvantage that the number of data matrices limits the separation result of RPCA algorithm.In addition,TV-TRPCA algorithm also breaks through the limitation those traditional multi-channel clutter suppression algorithms such as DPCA(Displaced Phase Center Antenna)and ATI(Along Track SAR Interferometry)need to meet the premise of geometric relationship.The experimental results show that the TV-TRPCA method realizes the effective suppression of clutter signals and shows its advantages in a variety of simulation environments.2.Aiming at the problem of moving target parameter estimation,this thesis uses the short-time of UAV platform echo signal and proposes matrix pencil MP algorithm based on single frequency signal model.Using the characteristics of hank matrix structure and the mathematical properties of single frequency signal,the parameter estimation problem of single frequency signal is transformed into the generalized eigenvalue problem of hank matrix.Using the mathematical properties of single frequency signal,the algorithm overcomes the insensitivity of traditional methods such as short-time Fourier transform(STFT)and fractional Fourier transform(Fr FT)to similar parameters in multi-component signals.PRONY algorithm also uses the mathematical properties of single frequency signal for parameter estimation.Compared with MP algorithm,the solution method is simple and the operation cost is reduced.The experimental results show the accuracy of the parameter estimation results of MP algorithm,and the estimation results are better than the traditional parameter estimation methods.