No abstract

No abstract

The ability to obtain reliable, long-term visual data in marine habitats has the potential to transform biological surveys of marine species. However, the underwater environment poses several challenges to visual monitoring: turbidity and light attenuation impede the range of optical sensors, biofouling clouds lenses and underwater housings, and marine species typically range over a large area, far outside of the range of a single camera sensor. Due to these factors, a continuously-recording or time-lapse visual sensor will not be gathering useful data the majority of the time, wasting battery life and filling limited onboard storage with useless images. These limitations make visual monitoring difficult in marine environments, but visual data is invaluable to biologists studying the behaviors and interactions of a species. This paper describes an acoustic-based, autonomous triggering approach to counter the current limitations of underwater visual sensing, and motivates the need for a distributed sensor network for underwater visual monitoring.

Acoustic methods are becoming increasingly common in the study of marine mammal populations and behavior. Automating the detection and classification of whale vocalizations has been a central aim of these methods. The focus has primarily been on intra-species detection and classification, however, humpback whale (Megaptera novaeangliae) social call detection and classification has largely remained a manual task in the bioacoustics community. To automate this process, we processed spectrograms of calls using PCA-based and connected-component-based methods, and derived features from relative power in the frequency bins of these spectrograms. We then used these features to train and test a supervised Hidden Markov Model (HMM) algorithm to investigate classification feasibility.