Font Size: a A A

Multi-species detection and classification of marine mammals

Posted on:2011-02-18Degree:M.ScType:Thesis
University:Royal Military College of Canada (Canada)Candidate:Marcotte, DavidFull Text:PDF
GTID:2440390002451995Subject:Physics
Abstract/Summary:
In recent years, there has been an increased awareness of the effects maritime operations (civilians and military) can have on marine mammal populations. In the past, one had to rely on the use of an experienced human operator (EHO) to acoustically detect such animals. The Canadian Forces are actively trying to develop automated ways to detect mammals in order to mitigate the danger associated with the use of powerful sonar and explosives in local population of marine mammals.;Different methods have been developed for the detection of mammal vocalizations: spectrogram correlation, hidden Markov model, matched filter, band limited energy detector and the fast orthogonal search (FOS). There are a wide variety of vocalizations: songs, moans, grunts, whistles, and echolocation clicks. This work is aimed at detecting echolocation clicks using a transient signal energy detector called the Teager-Kaiser Energy Operator. Two ways of classifying an echolocation click in different species of marine mammals have been developed.
Keywords/Search Tags:Marine mammals
Related items