Polarization Radar Target Recognition And Multi-classifier Fusion Research

Radar automatic target recognition technology based on high range resolution profile(HRRP)has achieved intensive attraction over the world for its advantages of high real-time reaction,easy access and coverage of important structure information of radar targets.This thesis is focused on the research of polarized radar target recognition based on HRRP and multi-classifier fusion.The main research contents and innovations are as follows:1.For broadband multi-polarized radar,this paper combines high range resolution profile with polarization information to obtain polarization distance matrix of target in four polarization configurations.The algorithm performs omnidirectional feature extraction and modeling on the target,which can be adapted to different poses,effectively reducing the influence of the HRRP azimuth sensitivity.At the same time,on the basis of polarimetric distance matrix directly,Pauli decomposition and Freeman decomposition are proposed to extract the target features of the polarization distance matrix,which paves the way for subsequent research.2.Deep convolution neural network is utilized to mine and recognize HRRP data with polarization information.Based on the polarimetric distance matrix directly,Pauli decomposition and Freeman decomposition are employed to extract the target feature of the polarimetric distance matrix,and the obtained target feature vectors are combined and sent to the deep convolutional neural network for training.This method not only combines different feature extraction methods to extract more features,but also learns the target feature vector with the application of deep convolution neural network.Simulation results presents the verification of this method.3.The fusion algorithm of multi-classifier radar target recognition based on the improved upper integral is studied.This method is able to reflect the importance of single base classifier and random combination of base classifiers,meanwhile it illustrates the interaction between classifiers.In theory,this method can ensure that the classification recognition rate of multi-classifier fusion system is not lower than that of any base classifier.Experiments show that this method can improve the rate of radar target recognition effectively.