To main content

QVision: Online quality detection of inhomogeneous food using spectral imaging

QVision: Online quality detection of inhomogeneous food using spectral imaging

Published 12 November 2014

A competitively-priced online spectral imager has been developed that is robust and accurate enough for industrial applications

Illustration: Tomra

SINTEF has a long-standing collaboration with Nofima and Tomra ASA, tackling the challenges with measuring food quality in several online industrial cases. Together we have developed QVision, a robust spectral instrument that penetrates and senses up to 20 mm into food.

The following industrial cases have been investigated:

  1. Fat content in salmon
  2. Fat content of inhomogeneous meat (pork and beef)
  3. Water content in salted cod
  4. Ice-fraction in superchilled salmon
  5. Meat-content in live crabs

The clients of Tomra have different demands, e.g. meat is valued higher than fat so the lower the fat content, the higher the selling value. Many quality parameters cannot be measured in real-time and online with the required accuracy. For example, salted cod was measured manually by bending the fish and getting a 'feel' for the resistance and relating it to the dryness. Another example is fat measurements, which can be done online using near infrared (NIR) spectroscopy but the food product often needs to be homogeneous. X-ray can be used on heterogeneous food but it is an expensive alternative. QVision is designed to be robust towards a reasonble amount of heterogeneity.
SINTEF and Nofima have, together with Tomra, developed a robust NIR optical measurement system combined with multi-variate analysis to achieve accuracies of better than 2% on inhomogeneous meat and better that 1% on homogeneous samples. A key success factor is employing what is often called transflection, interactance or elastic scatter spectroscopy in the literature. This means that the light penetrates and interacts with the food instead of just being reflected from the surface. More can be read about the system on Tomra's website where QVision is commercially available: https://www.tomra.com/en/solutions-and-products/sorting-solutions/food/process-analytics/qvision/

Other key project partners include:

  1. Animalia
  2. Nortura
  3. FACCSA
  4. Andreas Bjørge

Superchilled salmon, measured with QVision, with ice-fractions from 8.7% to 42.3%

Research Manager

Project duration

01/01/2005 - 01/12/2012

Projects funded by the Research Council Norway

  1. MeatVision
  2. MeatAutoSort
  3. Competitively-price food processing