When a radar tries to spot a floating object, like a boat or a wreck, the sea’s echo can drown out the target’s signal. In strong sea clutter conditions, waves generate a spectral noise that makes it hard to pick out the real echo. Traditional mode‑decomposition methods, which split the signal into simpler components, often fail to select the modes that actually contain the target.
The new approach presented in this paper uses a spectral mode reconstruction based on an adaptive selection criterion. First, it identifies the typical Doppler‑frequency interval of the target, then compares it with the statistical spectral features of sea clutter. Only modes falling within this “window” are reconstructed, effectively stripping away much of the noise.
To test its performance, the authors built an evaluation framework that measures relative feature gain and detection probability. Experiments on real measured data show that the reconstructed signals vastly outperform the original ones: background clutter is almost eliminated and target components become much clearer. This technique could boost maritime surveillance, making object detection in rough waters more reliable.
