Research On Ballistic Target Recognition Algorithm Based On Micro-doppler Effect

In recent years,with the vigorous development of decoy technology,traditional bal-listic target recognition based on the shape and infrared characteristics of the target has been less applied.At this time,the method of using the micro-motion characteristics of the target for target recognition has attracted the attention of scholars at home and abroad.The micro-motion characteristics of the ballistic target reflect the micro-motion form of the target and the characteristics of some components,which can reflect the difference between the warhead and the decoy in the ballistic target.Therefore,this thesis mainly studies the translational compensation algorithm of the mid-range ballistic target and the algorithm of extracting features from the micro-Doppler after translational compensation for ballistic target recognition.The main research contents are as follows:1.The micro-motion models of three ballistic targets of precession,swing and tum-bling are studied.With reference to the derivation process of the micro-Doppler frequency shift relations of precession and swing,the micro-Doppler frequency shift relations of tum-bling are deduced.The micro-Doppler features of the three ballistic targets are modeled and analyzed,which lays the foundation for the feature extraction of subsequent ballistic target recognition.2.Aiming at the problem that the extraction of the micro-motion feature of the mid-range ballistic target is affected by the translation,the impact of the translation on the micro-Doppler is analyzed,and the template method and the peak method are used to com-pensate the target with a single micro-movement.After analyzing the effects of translation compensation algorithms based on delayed conjugate multiplication and symmetric con-jugate multiplication,an improved peaking algorithm based on digital image processing was proposed.The comparison proves the superiority of the algorithm performance.Fi-nally,in the actual ballistic group target scene,a ballistic target translation compensation algorithm based on Radon transform center method is proposed,and numerical simulation proves that the algorithm is still at low signal-to-noise ratio(SNR).It has better transla-tional compensation performance.3.Aiming at the problem of identifying the mid-stage ballistic target,firstly extract the Hu invariant moment feature and the singular value feature of the quaternion matrix from the time-frequency image angle,and identify three types of ballistic targets: pre-cession warhead,swing decoy,and roll mother capsule.Simulation results proved that Hu invariant moment features has better recognition performance.Then three spectral line features and four time-frequency distribution features were extracted from the micro-Doppler domain,and three types of ballistic targets were identified.Finally,the recog-nition effect of the quaternion matrix singular value feature and the three features based on the frequency spectrum proposed in this thesis are compared.Numerical simulation proves that the spectral feature is better than the quaternion matrix singular value feature under low signal-to-noise ratio.