A Polarization-Doppler Joint Feature-Based Detection Method for Small Targets in Sea Clutter

The task of detecting small targets poses a significant challenge for ocean surveillance radars. Since the sea clutter is time-varying and the targets with small radar cross-sections (RCSs) have weak returns, traditional adaptive detectors that rely on the statistical model of clutter fail to work. An effective approach is feature-based detector, which primarily takes advantage of the feature difference between target and clutter. The discriminability of single-dimensional features such as amplitude and Doppler spectra is limited, while the potential of polarization information remains undeveloped. Stemming from the polarization dependence of the clutter Doppler spectrum, we propose an innovative polarization-Doppler joint feature-based detection method. Based on the polarization modulation that digitally synthesizes multiple transmitting and receiving polarization states, we construct a 2-D polarization-Doppler spectrum and extract the joint polarization-Doppler features, which can reveal the significant differences between clutter and targets. Subsequently, a detector based on the false alarm rate controllable support vector machine (SVM) classifier is proposed. The experimental results, derived from IPIX datasets, demonstrate that the proposed method exhibits enhanced robustness compared to traditional feature-based approaches. Furthermore, it has superior performance and less computational cost compared to the time-frequency feature-based method.