FMCW ISAR Imaging Technology Based On Convolutional Neural Networks

Inverse Synthetic Aperture Radar(ISAR)can perform high-resolution two-dimensional imaging on non-cooperative moving targets such as aircraft,missiles,ships and satellites.With the in-depth research of ISAR radar,the combination of Frequency Modulated Continuous Wave(FMCW)technology and ISAR technology makes the advantages of radar in various aspects prominent.FMCW ISAR not only has the advantages of simple structure,low price and fine quality,but also reduces the transmission power and intercept probability.However,compared with traditional ISAR imaging,FMCW ISAR imaging process is more complex,and some problems need to be further studied.In this paper,based on the principle and method of ISAR imaging,the complex motion compensation problem caused by target motion is studied.The following is the main content of this article:Firstly,the principle of traditional ISAR imaging is studied,and the scattering point model,radial motion and compensation of ISAR image are analyzed.And the relevant components of convolutional neural network CNN are briefly introduced.Secondly,an FMCW ISAR imaging method Based on Image Contrast Based Algorithm(ICBA)and OSS phase retrieval algorithm is proposed.Based on the FMCW ISAR radar and the target coordinate geometry model,the radial motion of the target in the stationary state is derived by the formula,and the phase error changes caused by the translation of the target and the errors to be compensated are obtained.For the phase errors caused by the radial acceleration of the target on the fast time,the ICBA compensation algorithm is used to carry out translational coarse compensation to complete the range focusing.For the residual phase errors caused by slow time and radial velocity,the OSS phase retrieval algorithm is used to carry out translational fine compensation to complete the azimuth focusing.The experimental results show that the proposed method can obtain better imaging results.Finally,an FMCW ISAR imaging method based on U-net network is proposed.Under the classical U-net network model,a residual module is added to improve the speed of network training.Construct FMCW ISAR algorithm flow under target motion state.The experimental simulation of the echo data after motion compensation under different conditions shows that U-net network can achieve high quality imaging of the target after motion compensation.