Logistic regression prediction based on fractal characteristics of sea surface targets

Abstract

Sea clutter is a kind of signal with low signal-to-noise ratio. Fractional Fourier transform is used to gather energy. Fractional Brownian motion model is introduced to model sea clutter. Based on the measured data of IPIX radar, high-order multi-fractal parameters are calculated. Fitting the multi-fractal parameters, choosing the steepest interval and calculating its slope. At the same time, choosing the fractal parameters with scale -30 under HH and VV polarization. The small targets are 0 and 1, respectively. The slope and high-scale fractal parameters of different sea conditions are normalized, and the normalized data are predicted by logistic regression. The simulation results show that FRFT can aggregate the energy of sea clutter signal. Logistic regression model can predict the fractal parameter data of sea clutter FRFT domain, and the prediction accuracy reaches 83.42%.