Analysis Of High Frequency Scattering Characteristics Of Radar Targets And Research On Its Recognition Application

Electromagnetic scattering characteristics of radar targets are of great significance in the field of radar target recognition.Effective use of radar target electromagnetic scattering information and analysis of target electromagnetic scattering characteristics can have a positive impact on the completion of radar target recognition tasks.The development of computer technology provides a good software and hardware foundation for the research of target electromagnetic scattering characteristics.At present,computer simulation technology has become one of the important means to accurately obtain target electromagnetic scattering information.At present,many computational electromagnetics methods can accurately predict the scattering characteristics of targets based on computer simulation technology.Among them,for electrically large size targets,high-frequency method requires low computer hardware foundation,fast calculation speed and high accuracy.Therefore,the high frequency method has been widely used in the simulation of high frequency scattering characteristics of electrically large radar targets.The simulation method of target scattering characteristics studied in this paper is based on high frequency method.Physical optics(PO)and geometric optics(GO)are used to obtain scattering field information of electrically large targets,calculate radar cross section(RCS)of targets and obtain synthetic aperture radar(SAR)images of targets.Aiming at target recognition,the research of radar target scattering center parametric modeling is further carried out,and scattering center parametric model is applied to template-based target recognition method and model-based target recognition method respectively.In the template-based target recognition method,the parametric modeling method can be used as a new method to obtain radar target feature data,and the neural network can learn knowledge which is beneficial to classifying the actually measured SAR images from the parametric model reconstruction SAR image data and help the neural network to complete the recognition task in the actually measured SAR image data.In the model-based target recognition method,the parametric modeling process associates the scattering information of the target with the geometric structure,and can obtain a clearer scattering mechanism of the target to complete the interpretation of the target component configuration.The main contents of this paper are as follows:1)The simulation method of radar target high frequency scattering characteristics is introduced.The PO method is used to calculate the primary scattering field of the target echo,and the bounce ray method(SBR)is used to obtain the secondary scattering field of the target echo.The scattering characteristics of simple scatterers and complex targets are simulated by using the high frequency scattering characteristics simulation method of radar targets,and the corresponding RCS characteristics and SAR image characteristics of targets are obtained.The scattering characteristics of the target obtained by the simulation method are analyzed,and the effectiveness of the method is verified.2)The parametric modeling method of radar target is introduced.Geometric models of multiple targets are established based on target components.On this basis,scattering centers for establishing parametric models are obtained by using ray diversity technology and scattering source selection method,various parameters required for attribute scattering center models are determined,and parametric models of multiple targets are constructed.The scattering mechanism of one of the targets is analyzed by using the parametric model of the targets,and the SAR images of multiple targets are reconstructed.The reconstructed SAR images are compared qualitatively and quantitatively with the SAR images obtained by the simulation method and the measured SAR images.3)Parameterized model is used in target recognition task based on template.The target SAR image is reconstructed by using the parametric model of the target,and the acquired SAR image is used to construct a "P-SAR" data set.The neural network is pretrained with the "P-SAR" data set,and the performance of the pre-trained neural network and the non-pre-trained neural network is compared.The experimental results show that the pre-trained neural network converges faster in the training process.4)The parameterized model is applied to the target recognition task based on the model.Geometric modeling is carried out on the target to be identified,information of scattering centers formed by specific components in the target geometric model is obtained forward through a parametric modeling method,and scattering center information is extracted backward from SAR images of the target to be identified.Matching the scattering center obtained in the forward direction with the scattering center extracted in the reverse direction to complete the component configuration interpretation of the target to be identified.