Feature-oriented striation processing of the noise field due to a moving ship

The regular striations exhibited in broadband noise spectrogram due to a moving ship have been exploited for inverse problems in underwater acoustics, e.g., sediment geoacoustic characterization and source ranging. The enhancement of the underlying interference structure for striation extraction has been a hot topic for relative applications. In this paper, a straightforward method based on the low-frequency sound propagation characteristics is proposed to strengthen the main striation features. This method intuitively enhances the striation pattern by smoothing the resultant spectrogram along range (time) for each frequency bin. The method is applied to the ship noise data collected in 2008, Dalian, and then filtered by a multi-scale line filter, which demonstrate can greatly improve the quality of extracted striations than that of raw data. A hough transformation technique is applied to reconstruct the striation segment to estimate precise geometrical information that can be used for source range estimation.