Bjørn Skovlund Dissing

Multispectral Imaging for Determination of Astaxanthin Concentration in Salmonids

Multispectral imaging has been evaluated for characterization of the concentration of a specific cartenoid pigment; astaxanthin. 59 fillets of rainbow trout, Oncorhynchus mykiss, were filleted and imaged using a rapid multispectral imaging device for quantitative analysis. The multispectral imaging device captures reflection properties in 19 distinct wavelength bands, prior to determination of the true concentration of astaxanthin. The samples ranged from 0.20 to 4.34 g per g fish. A PLSR model was calibrated to predict astaxanthin concentration from novel images, and showed good results with a RMSEP of 0.27. For comparison a similar model were built for normal color images, which yielded a RMSEP of 0.45. The acquisition speed of the multispectral imaging system and the accuracy of the PLSR model obtained suggest this method as a promising technique for rapid in-line estimation of astaxanthin concentration in rainbow trout fillets.

Temporal reflectance changes in vegetables

Quality control in the food industry is often performed by measuring various chemical compounds of the food involved. We propose an imaging concept for acquiring high quality multispectral images to evaluate changes of carrots and celeriac over a period of 14 days. Properties originating in the surface chemistry of vegetables may be captured in an integrating sphere illumination which enables the creation of detailed surface chemistry maps with a good combination of spectral and spatial resolutions. Prior to multispectral image recording, the vegetables were prefried and frozen at -30°C for four months. During the 14 days of image recording, the vegetables were kept at +5°C in refrigeration. In this period, surface changes and thereby reflectance properties were very subtle. To describe this small variation we employed advanced statistical techniques to search a large featurespace of variables extracted from the chemistry maps. The resulting components showed a change in both the carrot and celeriac samples. We were able to deduct from the resulting components that oxidation caused the changes over time.

Quality assessment of butter cookies applying multispectral imaging

A method for characterization of butter cookie quality by assessing the surface browning and water content using multispectral images is presented. Based on evaluations of the browning of butter cookies, cookies were manually divided into groups. From this categorization, reference values were calculated for a statistical prediction model correlating multispectral images with a browning score. The browning score is calculated as a function of oven temperature and baking time. It is presented as a quadratic response surface. The investigated process window was the intervals 4–16 min and 160–200°C in a forced convection electrically heated oven. In addition to the browning score, a model for predicting the average water content based on the same images is presented. This shows how multispectral images of butter cookies may be used for the assessment of different quality parameters. Statistical analysis showed that the most significant wavelengths for browning predictions were in the interval 400–700 nm and the wavelengths significant for water prediction were primarily located in the near-infrared spectrum. The water prediction model was found to correctly estimate the average water content with an absolute error of 0.22%. From the images it was also possible to follow the browning and drying propagation from the cookie edge toward the center.

Spectral Imaging as a Tool in Food Research and Quality Monitoring of Food Production

A forward-looking food industry must obviously continue to develop its production technology to be able to produce foods that meet both present legislation and consumers’ expectations and demands. Ensuring a healthy, secure and sensory food quality, as well as ensuring cost competitiveness / effectiveness is of high importance to survive the strong competition within the field. The high costs in many food-processing areas are primary caused by the extensive use of manual work e.g. for visual inspection of quality parameters and the subsequent sorting or removal of products. However, new production and / or distribution technologies in themselves neither create nor ensure high quality of products or the optimization of the production. This requires knowledge about both new opportunities to create specific production and distribution conditions in combination with knowledge about product response given the production and distribution conditions. Food and food production covers a broad variety of both raw materials and production processes. Therefore, application of new technology cannot be regarded as a simple procurement of accessible standard products but requires research and development including several tests to ensure optimal outcome. The assessment of the visual appearance of food products from size and colour of the product to uniformity of packaging is an important part of the control system in the food supply chain. To ensure that the required standards are met, inspections at all stages from primary production to final retail distribution are needed. However, visual inspection of quality parameters and the subsequent sorting and sometimes also rejection of products by manual work are significant contributors to the total production costs in the food industry. To save costs and to enhance visual quality assessments, automatic vision systems are introduced and tested in many food-manufacturing operations. Vision systems are attractive for online quality assessment and process control, because the methods are rapid, contact free and non-destructive. The introduction of vision systems for quality monitoring in the food manufacturing industry is challenged by the relatively harsh production environment, typically in the form of a high humidity, low / high temperatures, routine wash down and sanitation. The technology is maturing, however, and the special hardware requirements on system design are now met, so that computer-based vision systems are now gaining wider application for quality monitoring in food processing.