Research On Automatic Recognition Of Ship Targets Based On The Combination Of HRRP And ISAR Image

In the context of information warfare,modern radar can not only provide the position and speed information of the target,but also image the target to obtain the geometry and attitude information of the target,and the automatic target recognition technology of the radar also follows.Among them,due to my countrys complicated marine disputes,the identification of ship targets is of great significance to my countrys territorial security.This article focuses on the research of radar ship target recognition.Due to the current research on the use of HRRP or ISAR images for recognition,this paper mainly focuses on the fusion of HRRP and ISAR images for ship target recognition.Research,including the method of fusing the two for target recognition to improve the recognition rate and the method of fusing the two to improve the accuracy of length estimation.As for the fusion recognition method,this paper adopts fusion based on DS evidence theory and fusion based on classifier selection.In terms of length fusion,from simple to complex,the zero-mean Gaussian distribution and the non-zero-mean Gaussian distribution are used to model the length estimation error of HRRP and ISAR.Use the training set data to estimate the parameters of the Gaussian distribution,and then use the estimated probability model to pass the minimum mean square error estimation method,and the HRRP and ISAR length estimation results are fused.The ISAR image length estimation results are different from the HRRP length estimation results.For large samples,the HRRP length estimation result is directly taken as the fusion result.The verification results on the radar imaging simulation data set show that the fusion method based on the classifier selection can get the best accuracy and the accuracy of the length estimation after fusion has been significantly improved.First of all,in order to get the simulation data set,this article describes the principle of ISAR imaging,and the method of modeling the scattering point and the motion state of ship targets.Simultaneously,the K distribution model was used to simulate sea clutter,so as to obtain simulated HRRP data and ISAR image data with the influence of sea clutter,which were used to evaluate the performance of the processing method.Before the fusion process,for the HRRP length estimation and feature extraction,CFAR detection is used to obtain the range of the target in the distance dimension,and the estimated value of the target length is obtained.In feature extraction,bispectrum features with constant translation are used.In the length estimation and feature extraction of ISAR images,in addition to the traditional CFAR method for target area extraction,this paper also proposes a method for target area extraction from the perspective of image segmentation using U-Net network.Finally,the fusion part uses the aforementioned fusion method to process on the simulation data set to verify the performance of the algorithm.The results show that the recognition accuracy of the classification decision fusion method proposed in this paper has improved,and the error of the length estimation fusion result has also decreased,indicating the effectiveness of this method.