Acoustic species identification in the Northwest Atlantic using digital image processing

Abstract

Acoustic surveys for marine fish in coastal waters typically involve identification of several species groups. Incorrect classification can limit the usefulness of both distribution and biomass estimates. Fishing catch data can assist in identification but are rarely spatially comparable to acoustic data and are typically biased by gear type. We have developed analytical tools to enable identification of Atlantic cod ( Gadus morhua) and capelin ( Mallotus villosus) using high-resolution echograms. The approach is to assess and analyze various features of the acoustic returns from shoals and individuals using image processing techniques, then to use these features in a learning mode to develop algorithms that discriminate among species. A Mahalanobis distance classifier, which uses the covariance matrix for each species in its distance measurement between species, has been implemented and tested. We demonstrate these techniques using the software “FASIT”, developed for that purpose, in the analysis of inshore fisheries data from Placentia Bay, Newfoundland using data from a 38 kHz digital echo sounder.