Automated two-dimensional localization of underwater acoustic transient impulses using vector sensor image processing...

Area of Science:

• Marine bioacoustics
• Underwater acoustics
• Signal processing

Background:

• Detecting acoustic transients using signal-to-noise ratio (SNR) is difficult in nonstationary ambient noise, common in coral reef ecosystems.
• Traditional passive acoustic detection methods struggle in complex underwater soundscapes.

Purpose of the Study:

• To develop and validate an automated algorithm for detecting and localizing transient impulsive sounds in challenging underwater environments.
• To utilize signal directionality as a primary criterion for acoustic event detection, independent of precise time synchronization.

Main Methods:

• Constructing time-frequency representations (spectrograms and azigrams) from underwater vector sensor data.
• Employing binary image processing to link time-frequency cells across sensors based on shared duration and bandwidth.
• Triangulating sound source locations using identified directional acoustic events.

Main Results:

• The algorithm successfully detected SCUBA bubble plumes and fish activity in shallow coral reef environments.
• False transient localization rates were estimated between 3% and 6% in the studied coral reef environment.
• Localized acoustic pulses exhibited a median SNR of 7.7 dB with an interquartile range of 7.1 dB.

Conclusions:

• Signal directionality offers a robust alternative to SNR for detecting acoustic transients in noisy, nonstationary environments.
• The developed algorithm provides accurate localization of underwater acoustic sources, demonstrating its utility in marine ecological studies.