"One Fish, Two Fish, Red Fish, Blue Fish"
This famous line from Dr. Seuss is about to take on a whole new meaning for a team of William and Mary scientists. VIMS researchers Mark Patterson and Roger Mann, along with Dr. Zia-ur Rahman of the College's Department of Computer Science, are combining high-resolution side-scan sonar, sophisticated image analysis, and robotics to identify and count fish.
The research also involves VIMS graduate student Daniel Doolittle and partners in the marine technology industry.
The team has mounted a sonar device on a miniature submarine called an Autonomous Underwater Vehicle, or AUV. The sonar unit, which was developed by a team led by Marty Wilcox of Marine Sonic Technology Ltd. in Gloucester, resembles the "fish finders" used by anglers, but provides images of such high resolution that the shapes of individual fish can be clearly discerned. The unit produces images by emitting sound waves and recording the echoes produced when the waves reflect off underwater objects. The remotely controlled AUV, developed by Sias Patterson, Inc., can dive to depths of 1,000 ft and stay submerged up to 4 hours.
Fetch current specifications
Length: 1.73 m |
With the support and cooperation of the Virginia Aquarium in Virginia Beach, the team is beginning their work in the facility's Norfolk Canyon and Chesapeake Bay aquariums by gathering images of selected fish species. Next, they will "train" a computer so that it can, when fed a digital sonar image, quickly count and measure individual fish of selected species based on their shape. To ease the computer's task, the team plans to enhance the sonar images prior to analysis with a state-of-the-art image-processing system developed by Dr Rahman. This includes adjusting the image so that the shape analysis is not confounded by the orientation of the fish relative to the sonar unit.
Marty Wilcox of Marine Sonic Technology, Ltd, says, "We have observed single fish and schools of fish while developing and using our high-resolution sonar, but this is taking the technology in a whole new direction. We are very excited to be a part of this very innovative and ultimately very beneficial application of our technology."
To effectively manage commercial and recreational fisheries, regulatory agencies need to know the populations of different fish species and how they vary through time. But accurately counting fish isn't easy. Traditional methods, in which scientists count fish caught in a net or on a line, only provide a "snapshot" of a population at a specific time and place. Such data are less than ideal, particularly in estuaries like Chesapeake Bay, where fish populations vary tremendously from day to day, year to year, and place to place.
If the team is successful, assessment of fish stocks in the Chesapeake Bay and other estuaries will never be the same, as a fleet of instrumented AUVs augment the use of trawl nets and long lines, and help to meet the growing need for improved fisheries data.