(July 23, 1998, Gazette)
Dr. Gosine said increasing the reliability of methods used for fish fillet inspection has been an impediment for the fish processing industry for a long time. But thanks to the Parasensor, this problem may be solved once and for all.
"To date, there isn't any equipment commercially available and nobody has shown any techniques to solve this particular problem."
In a recent interview with the Gazette, Dr. Gosine explained that although the fish processing industry has become highly automated, one aspect of the process - fish inspection - is still done manually. He said visually inspecting fish is extremely labor intensive, costly, and can limit a fish processor in providing the highest quality product possible.
"If Canada is to have a competitive fish processing industry, it is essential to have highly reliable methods for detecting defects in such products," said Dr. Gosine of Memorial's Engineering Faculty. "We need to provide economic solutions to very challenging technical problems associated with this objective."
Dr. Gosine believes the Parasensor prototype is the answer to the fish inspection problem. The advanced image-analysis technology automatically inspects fillets as they move along conveyor belts. Images of the fish are captured by a video scanner, and the images are converted to digital signals. These signals are then analysed by innovative software, which distinguishes the fish defects and automatically sorts out the affected fillets.
Dr. Gosine noted that the Parasensor can analyse the fish at a rate of one fillet per second and make decisions about the product in an instant.
"The objective of the Parasensor project is to develop reliable automated visual inspection techniques that can very rapidly determine that a fillet is free of defects so that those fillets do not need to be removed from the processing line, which is already highly automated."
The Parasenor system is presently undergoing full-scale testing involving half-a-million fillets at a fish plant in Arnold's Cove. Testing will continue throughout the summer to determine how well it performs under complete industrial conditions.
Dr. Gosine has been doing research on the defect detection system on behalf of C-CORE for the past two years. The project is a collaboration between a number of groups, with Canpolar East Inc. of St. John's being the prime contractor and doing the engineering work for the system. The main sponsors are PRECARN Associates and the Canadian Space Agency. Other collaborators include the Centre for Aqauculture and Seafood Development at the Marine Institute, the Fisheries Council of Canada and the Canadian Centre for Fisheries Innovation. The University of British Columbia, Concordia University, Spar Aerospace and MacDonald Dettwiler are technical advisors for the project.
Dr. Gosine said the Parasensor is a critical step forward for the fish processing industry, and added that when testing finishes in the fall, a new product will be available that he believes could have a major impact on the industry.
"Although we have been out of the business of producing cod for several years, when the market rebuilds you have to come back with a premium product. The best possible product has to be brought back into the market at a good price.
"On one level, this solution to the problem is essential because if you can't be sure of producing and guaranteeing a high quality product then who's going to want to give up steak and go back to fish?"
He said the Parasensor is important technology and a key element in the success of a fishing industry.
"Everybody who lives here must have some concern of the impact of our downturn in the fishing industry ... when the fish come back, the processors have to come back with the best quality product and this technology is going to be one step to ensuring that happens.
"I think everyone can be convinced that we need a fishing industry or that we miss not having one, so this technology is going to be one of the tools that will mean we will have an industry. That is important to everyone."
Although the immediate importance of the Parasensor is to the fish processing industry, Dr. Gosine said it can also be applied to many other things.
"These techniques that we have developed are general enough that they can be trained to recognize other things, like wood detection or even inspecting cars that are on the highway by zooming in on the license plate to capture information about a car."
He said the new technology can be applied to our mining and forestry industries, adding that one of our failures in Canada has been our inability to develop the technologies to support our resource industries.
"While the mining, forestry and fishery employ very sophisticated technology, most of that technology is imported from Europe. The Parasensor project is an example of very advanced technology that has been developed in Canada to support our resource industry, but which will have significant international applications."
Dr. Gosine said his participation in this research project has been gratifying and enjoyable and he pointed out that three other C-CORE researchers - Patricia LeFeuvre, Rod Hale and Dr. Charles Randell - are to be commended for their contributions to the project as well.
"This is one of those projects that you feel is very worthwhile doing because you have a sense of how important it is locally. And when you look around at the impact of the downsizing of the fishery, if there's a way to contribute to the technology that will help maintain that industry, then it's satisfying."