Bayesian Detection and Tracking of Odontocetes in 3-D from Their Echolocation Clicks

TL;DR

This paper introduces a novel automated method for detecting and tracking marine mammals in 3-D space using echolocation clicks and TDOA measurements, improving accuracy over previous manual approaches.

Contribution

It presents a fully automated, graph-based multi-target tracking system that processes TDOA data for 3-D localization of odontocetes from volumetric hydrophone recordings.

Findings

The method successfully tracks Cuvier's beaked whales in real acoustic data.

Simulation results show the approach outperforms manual annotation methods.

The system effectively rejects false detections and closes detection gaps.

Abstract

Localizing and tracking of marine mammals can reveal key insights into behaviors underwater that otherwise would remain unexplored. A promising nonintrusive approach to obtaining location information of marine mammals is based on recordings of bio-acoustic signals by volumetric hydrophone arrays. Time-difference-of-arrival (TDOA) measurements of echolocation clicks emitted by odontocetes can be extracted and used for detection, localization, and tracking in 3-D. We propose a data processing chain that automatically detects and tracks multiple odontocetes in 3-D from their echolocation clicks. First, TDOA measurements are extracted with a generalized cross-correlation that whitens received acoustic signals based on instrument noise statistics. Subsequently, odontocetes are tracked in the TDOA domain using a graph-based multi-target tracking (MTT) method to reject false TDOA measurements and close gaps of missed detections. The resulting TDOA estimates are then used by another graph-based MTT stage that estimates odontocete tracks in 3-D. The tracking capability of the proposed data processing chain is demonstrated on real acoustic data provided by two volumetric hydrophone arrays that recorded echolocation clicks from Cuvier's beaked whales (Ziphius cavirostris). Simulation results show that the presented 3-D MTT method can outperform an existing approach that relies on hand annotation.