Research On Radar Target Detection Method Based On Time-frequency Manifold

The detection of sea-surface weak targets has always been a difficult problem in the background of sea clutter.In this paper,a novel signal representation method based on time-frequency analysis and information geometrical theory is studied for weak target detection in sea clutter background,and a time-frequency manifold including time-frequency geometry is proposed.The difference measurement and mean estimation method on time-frequency manifolds are studied,and the radar target detection method based on time-frequency manifolds is proposed.The framework of target detection method based on time-frequency manifold is presented in this paper,which provides a new idea for radar target detection based on information geometry theory.The main contents of this paper are as follows:The first chapter introduces the existing methods of detecting weak targets at sea surface,and compares the advantages and disadvantages of different features in detecting targets at sea surface.In addition,this chapter introduces the research status of radar target detection method based on information geometry,and summarizes the advantage of radar target geometry detection method under the condition of short observation time.The second chapter introduces the micro-Doppler model of the sea surface target,which provides the basis for the simulation of the target.At the same time,the method of radar target geometric detection is introduced,which lays a theoretical foundation for the study of matrix detection method.In the third chapter,the time-frequency manifolds are constructed to characterize the sea clutter and target signals,which combine the time-frequency characteristics based on time-frequency analysis and the geometrical characteristics based on information geometrical theory.In the time-frequency manifolds,time-frequency correlation is used to characterize the time-frequency distribution,and based on this,the time-frequency correlation difference measurement on time-frequency manifolds is studied.Aiming at the problem that the sea spikes have great influence when a few reference elements estimate the mean value matrix,the method of estimating the mean value matrix of Gaussian kernel function is put forward,and the influence of sea spikes in the mean value estimation is reduced,which provides a new idea for estimating the mean value matrix more accurately.In order to reduce the false alarm rate,the normalized time-frequency correlation matrix detection method is proposed on the time-frequency manifold to provide a new method for detecting faint targets at sea surface.In the forth chapter,we study the difference measurement method based on Moore Penrose pseudo-inverse on time-frequency manifolds,and propose the pseudo-inverse measurement and symmetric pseudo-inverse measurement.Based on this,the mean value matrix estimator method based on pseudo-inverse and symmetric pseudo-inverse in time-frequency manifold is studied,and the validity of Gaussian kernel mean estimator method in the previous chapter is verified,which provides theoretical basis for detector design.Aiming at the long estimation time of mean value of pseudo-inverse measurement,this paper studies the essence of difference measurement of pseudo-inverse measurement,proposes singular value difference measurement,and based on this,proposes singular value matrix detection method,which provides a new basis for the analysis of geometrical structure on time-frequency manifolds.The fifth chapter summarizes the main contents and innovations of this paper,and summarizes the shortcomings of the research and the prospects for the future.The research in this paper provides a new theoretical basis for the geometrical detection of sea surface targets,provide a new technical approach for the time-frequency detection of sea surface targets,and provide a new idea for the research of non-stationary signal processing fields.