Scientists at the Foundation for Scientific and Industrial Research (Sintef) found during tests that the computer vision system and the panel of inspectors evaluated the samples almost identically.

The colour of salmon is one of the most important quality parameters in processing because consumers associate redder fish with quality, taste and freshness.

Generally, the grading of salmon fillets is performed manually by trained inspectors using the internationally recognised standard of Roche colour cards. While vision systems are commonly used for sorting and quality control in many food industries, the uptake has been slower in fish processing.

However, previous studies estimate that replacing labour with vision systems, operating at a hypothetical rate one fillet per second, would reduce costs by €0.75 ($1) per kilo of fish product, based on current Norwegian processing practices.

The tests used adult fish from two processing plants in Norway. The fish were filleted and chilled, before being examined by the computer vision system using a Minolta chromameter, and a panel of three inspectors using Roche cards in daylight.

According to Norwegian industry standards, a "superior grade" fish is without substantial faults, damage, or defects, and provides a positive overall impression.

The test results showed almost no difference between the assessment of salmon quality, based on colour, by the computer vision system and that of the sensory evaluation using the Roche cards.

This demonstrates that a method based on computer vision to classify fillets can work, claims the Sintef.

"This method was a fast, nondestructive, and contact-free evaluation and was not significantly different from the traditional method of evaluating the colour by human vision," the Sintef said. "The better side of the computer vision method is that this method is faster, robust, and consistent."

Using a vision system could also reduce the need for operator facilities, lighting, heating, and other processing requirements.

The scientists also claim another advantage of the vision system is that it can scan a whole fish and produce a mean assessment, or examine specific areas of the fillet. Furthermore, the non-contact aspect of inspection also reduces the risk of contamination, Sintef claims.

Sintef is an independent research organisation that supports the development of about 2000 Norwegian and overseas companies. It main operations are in Norway, with further offices across Europe and the US.