Research On Radar Target Detection Method Under Complex Clutter Background Based On Probabilistic Statistical Model

Research on radar target detection technology in homogeneous and stationary clutter environment has become increasingly mature.However,in the background environment clutter is often non-uniform and non-stationary.Therefore,modeling the clutter into a single distribution does not meet the conditions of CFAR,and there is no guarantee that the reference cells and the detected cells are subject to be the same distribution,resulting in a decrease in detection performance.The important part of this paper focuses on the signal model and the traditional methods of target detection,and studies on the clutter modeling and target detection in complex background.The main contents of this paper are as follows:1.This paper starts with the signal model and analyzes the target fluctuation model,noise model and clutter model.Then the four Swerling models and their applicable conditions are introduced.For the clutter model,several common clutter models and their applicable environments are analyzed.Then the the classic threshold detection is deduced by the optimization method,the necessity of CFAR detection is introduced by classic threshold detection.And The parameters of the CA-CFAR are derived and their advantages and disadvantages are analyzed.2.Then Fit the measured data with a traditional clutter model and analysis of the mismatch problem of traditional methods.Then the distributions of different types of clutter are compared,and analyzing their differences.For complex clutter scenes,using the Doppler spectrum of clutter as a feature to learn the probabilistic statistical model for learning clutter.This paper characterize different clutter backgrounds using a mixed Gaussian model.By modeling the clutter background,on one hand,we can classify clutter and obtain independent and identically distributed clutter regions.On the other hand,it can aslo be used as a means to effectively detect targets.3.The last,comparing the similarities of clutter cells at different locations with similar backgrounds by calculating the Kullback-Leibler(KL)distance,to verify sharing sample through homogeneous regions is feasible.For the small homogeneous region of detection cell and the stability condition of the reference cells of the surrounding environment is difficult to meet,resulting in a problem of insufficient sample and low accuracy of parameter estimation.Borrow independent and same distribution clutter cells aiming at background uniformity detection.For the new data collected continuously,the parameters of the cluuter model are updated in real time due to the degeneration of the clutter.Make the clutter model more accurate description of each frame of data,which may further improve detection performance.