Research On Ballistic Target Recognition Based On ISAR Image

As a new type of ballistic target recognition method,Inverse Synthetic Aperture Radar(ISAR)image,can provide detailed information related to the geometric structure and micro-motion characteristics of ballistic targets.It has very important military significance and long-term development prospects in the field of ballistic missile defense.This thesis aims to study in depth the problem of ballistic target feature extraction and recognition based on the ISAR image sequence of ballistic targets in the context of midrange ballistic target group recognition.The specific work of the thesis is as follows:1.The micro-motion characteristics of ballistic target are investigated,in combination with the scattering point model and the micro-motion model.By establishing the ISAR imaging model of ballistic targets,the influence of standard motion compensation on the imaging quality is analyzed.On this basis,the ISAR imaging process of ballistic targets based on time-frequency analysis is discussed as the basis of subsequent research.2.Aiming at the problem of parameter estimation accuracy of ballistic target under low signal-to-noise ratio,based on ISAR image sequence,this thesis firstly proposes a scattering point correlation algorithm based on Markov nearest neighbor correction to update the cross-correlation points,which effectively solves the association error problem of traditional nearest neighbor method when the distribution of scattering points is close,and improves the accuracy of micro-motion projection curve extraction.Secondly,in terms of improving the signal-to-noise ratio of the estimation interval and selecting a high-precision frequency estimation method,a Root-MUSIC frequency estimation method based on refined domain search matching is proposed.Compared with the traditional FFT estimation in the entire micro-motion projection interval,it effectively improves the frequency estimation accuracy when the micro-motion frequency is not an integer multiple of the frequency resolution under low SNR.Finally,a structural parameter estimation algorithm based on AFSA-PSO step-by-step optimization is proposed,which effectively addresses the issue that traditional PSO algorithms are prone to falling into local optimum when optimizing high-dimensional cost function,and ensures the global optimality of parameter estimation.The experimental results show that the estimation accuracy of precession frequency and nutation frequency in low SNR scenario is higher than that of traditional FFT estimation,and the mean square estimation error of cone height,precession angle and nutation angle is better than that of traditional PSO algorithm.3.Aiming at the problem of low recognition efficiency and low recognition accuracy of traditional BP neural network for ballistic target recognition,this thesis proposes a ballistic target ISAR image recognition recognition algorithm based on CCPSO-BP neural network and micro-motion domain division.In terms of recognition accuracy,the weights and thresholds of the BP neural network are optimized using the CCPSO algorithm,and maintains the diversity of particle optimization direction in traditional PSO algorithm through the ergodicity of composite chaotic map,so as to enhance the global search ability.In terms of recognition efficiency,the algorithm combines the division and selection of sub-micro-motion domains to reduce the amount of training data in the model and accelerate the speed of model learning.According to the simulation results,the proposed algorithm enhances the speed of convergence and the capacity to eliminate local optimum when compared to the traditional BP algorithm,and further improves the recognition accuracy while ensuring higher recognition efficiency.In summary,this thesis proposes an ISAR image ballistic target recognition algorithm based on improved BP neural network with micro-motion domain division,as well as an ISAR image ballistic target micro-motion and structural feature extraction and recognition algorithm based on ISAR image sequence,both of which effectively improve the recognition accuracy and efficiency of ballistic targets and provide a theoretical basis for ISAR imaging recognition of ballistic targets.