Click to learn more about the other Massachusetts Sea Grant program!
May 3, 2012
Georges Bank seafloor image processing. Brown Bag Lunch with Seth Newburg and Michael Soroka: AUV Lab, MIT Sea Grant
Date: Tuesday, May 29, 2012
Bring your lunch and learn about monitoring for invasive species with a MITSG autonomous underwater vehicle
An aggressively growing sea squirt, Didemnum vexillum, is found growing on Georges Bank and throughout coastal subtidal areas from Connecticut to Maine. The sea squirt grows over sessile organisms and often is the most abundant species in the community where it is found. On Georges Bank it may impact fisheries by restricting groundfish feeding (Valentine, et al. 2007) and threaten the scallop fishery by preventing scallop settlement (Morris, et al. 2009).
In order to document impacts and predict dispersion and distribution of Didemnum, we need to quickly survey the sea floor. To date, only a small portion of Georges Bank has been surveyed using cameras and videos. A system is needed that can cover a large area and identify potential Didemnum infestations in order to document the scope of the invasion. Autonomous underwater vehicles (AUVs) offer a platform for testing sensors, comparing the records to digital images, and collecting samples for verification.
Capable of untethered dives, the Odyssey IV AUV collects high resolution images from 1 m altitude off the sea floor. A CTD sonde mounted on the vehicle collects environmental data, which is time stamped and stored in an onboard database with the images. The Odyssey IV cruises at speeds up to 1.4 m/s and hovers using vectored thrusters. The Odyssey IV AUV is a valuable tool for surveying and detecting the presence of Didemnum at low cost and without causing damage to the benthic habitat.
Some of the first deployments of the Odyssey IV were to perform bottom-imaging surveys on Georges Bank, and 2700 images of the seafloor were captured. These image sets revealed the presence of Didemnum, however, photographs made by the deep water imaging system lack the contrast and color fidelity needed to accurately identify and study the marine organisms. Image processing algorithms were developed and used to compensate for imaging artifacts and enhance the photographs in order to reveal the biodiversity in this marine habitat. It was found that creating mosaics of overlapping photographs using the SIFT-RANSAC method (Brown and Lowe, 2007) is helpful in collecting census data of the benthic community.
About Seth Newburg
About Michael Soroka
About the Brown Bag Lunch Series