Research On Sea Surface Target Detection And Classification Technology Based On FRFT Domain Features

Radar target detection algorithms are of great significance in various fields such as military and civilian.The detection performance is related to China's defense territory and sea security.Due to the non-stationary,non-uniform,and non-Gaussian characteristics of sea clutter under highresolution radar,traditional target detection methods result in the low detection probability and the high false alarm rate,when sea clutter is modeled.Thus,it is difficult to detect small target in sea clutter.Based on the time-frequency characteristics of measured sea clutter data,the fractional Fourier transform is suitable for processing such signals.So the three-feature joint detection in the FRFT based on the fast convex hull learning,the fractal feature analysis in the FRFT,and threefeature SVM classification are proposed in the thesis.The research of this thesis is as follows:The definition and time-frequency characteristics of FRFT are given,as well as the properties and characteristics.A two-dimensional peak search method is introduced to,effectively obtain the optimal rotation angle.The fractal theory is also introduced,and it is verified that the incremental sea clutter can be modeled as the fractional Brownian motion,laying a good foundation for detection.FRFT is used to detect target based on the different energy concentration between sea clutter and target.The information entropy,peak-to-average ratio and peak standard deviation in the transform domain are extracted,respectively.The fast convex hull learning algorithm is used to obtain the detection decision region in the three-dimensional feature space,and a three-feature joint detection is proposed.Compared with the single feature detection,the proposed three-feature joint detection can obtain better detection performance in a short observation time.The characteristics of the sea clutter were analyzed by the Detrended Fluctuation Analysis.The single-scale fractal of the sea clutter is studied.An appropriate scale-free interval is selected,and the accurate fractal dimension and the variance of fractal dimension are extracted.The proposed two-feature detector effectively distinguishes the target from the clutter.The SVM is introduced.The three features,the information entropy,the peak-to-average ratio,and the peak standard deviation proposed in Chapter 3,are trained.The sea clutter is classified and predicted through simulation experiments,The SVM has a high accuracy rate,and can classify targets and clutter effectively.