| Inverse Synthetic Aperture Radar(ISAR)imaging technology uses the relative motion between the target and the radar to form high-resolution images.ISAR can not only overcome bad weather such as fog and snow,but also works at night without natural light.It plays an important role in military fields such as target identification,discrimination and classification.When the target has a simple motion,the traditional imaging algorithm can obtain a high-resolution radar image,but when the target has a complex motion,it is difficult to reconstruct a high-quality image by the traditional algorithm.Based on this,an ISAR self-focusing imaging algorithm based on deep learning is proposed.Specific research contents are as follows:Firstly,the geometric model of ISAR imaging was established,focusing on the basic concepts of ISAR imaging,such as one-dimensional range-orientation and azimuthal-orientation,translational compensation and rotational compensation,Range Doppler(R-D)imaging algorithm and ISAR imaging algorithm based on Oversampling Smoothness(OSS)were introduced.The two algorithms were analyzed through experimental simulation.Secondly,in view of the large angle motion target,the maneuvering target imaging geometry model was established.In this paper,a motion compensation algorithm based on the keystone transformation and deep learning was proposed,we adjusted and optimized the network structure which was based on the characteristics of ISAR imaging of u-net.We also use keystone transform to compensate the Range Cell Migration(RCM)caused by the target rotation movement.The training set,which contains 3000 samples,is obtained by the simulation experiments.After training,the network can improve the ISAR imaging resolution.The test results of point target and surface target under different motion parameters were wonderful,which proved the effectiveness of the method.Finally,in view of the sports complex maneuvering target,we put forward a method based on Gabor Wavelet Transform(GWT)and Convolutional Neural Networks(CNN)of ISAR imaging algorithm.The GWT was used in the motion compensation and ISAR imaging as a time-frequency analysis tool.GWT eliminated the doppler shift of time-varying,and it had rough compensation effect on motion target.Then,a clear ISAR image is obtained by using CNN for high resolution imaging. |
