Small Target Detection in Sea-Clutter Based on Time-Doppler Spectrum

In this paper, a small target detection method on sea-surface based on Time Doppler spectrum (TDS) is proposed. Under HH polarization, the TDS of clutter bin and clutter with target bin reflects different frequency shift energy distribution. Each energy distribution patch can be a feature of clutter and clutter with target classification. The convolutional neural network (CNN) has the function of automatically extracting image features to avoid the incomplete description of patches by manual features. As a result, a detector using CNN to extract TDS features is designed. The samples of bin under test (BUT) is the unknow-classification data set of inputting the detector to test its performance. The experimental results obtained from measured data of IPIX radar show that the detector can achieve more than 90% accuracy of small target detection in sea-clutter.