A Ship Detection Algorithm Based on Truncated Statistics

Abstract:
A new constant false alarm rate detector is proposed for ship detection in single-look and multilook intensity synthetic aperture radar images. The method is aimed at multiple target situations where the sea clutter statistics are estimated from a sample which is potentially contaminated by targets. It uses truncation to exclude outliers from the sample and truncated statistics to analyse the truncated sample in a statistically rigorous manner. Experiments show that the detector performs on par with state-of-the-art methods at lower computational cost, has excellent false alarm regulation properties, and can estimate sea clutter statistics from a window centered at the cell under test.