Project Hyperspectral Cameras

Printer-friendly versionSend by email

In the framework of the Intelligence For Food project the spectral camera systems developed by  IMEC and the research group MeBioS (KU Leuven) were identified as promising innovative sensor systems for the detection of foreign objects, latent defects, mycotoxine producing moulds, for thorough quality inspection of product surfaces as well as for visualization of product composition (water, sugar, fat, protein). This technique also offers possibilities for hygiene screening as some invisible aspect scan be ‘visualized’. Hyperspectral camera’s are the vision systems of the future as the can extract faster more detailed information from one image and even from one pixel. They combine traditional image processing with spectroscopy . They do not only use the three broad wavelengths of the visible light (the three basic colours red, green, blue) and the broad NIR (near infra red) domain. For every pixel of the image, the light is split into many narrow wavelength bands, resulting in a particular spectral fingerprint of an object: a unique spectral curve giving very detailed information about that object. The higher the amount of wavelength bands, the higher the spectral resolution.

This project aims at making the technology of spectral camera’s accessible for the Flemish food industry. A prototype is yet available the will be used for demonstrations, test under industrial conditions, further knowledge development, optimization research and validation in collaboration with the participating companies. The validation project combines technical evaluations (sensitivity, detection limit, robustness, measuring rate,…) with considerations on the balance between cost and benefits. Knowledge transfer concerning fast quality assessment with spectral cameras in the food industry is an important aspect within the project. Support for applications in the VNIR domain (400-1000 nm) will be provided by IMEC, while MeBioS (K.U.Leuven) will support the NIR based applications (1000 – 2500 nm). Taking into consideration costs and benefits, it will be determined which system can be implemented in the company; a flexible but high-cost spectral camera or a low-cost vision system based in standard camera’s and LED illumination on specific wavelengths. Using line scanners this technology is suited to be implemented above transportation belts or sorting lines. A complete system architecture will be developed that can be elaborated by an integrator (3rdparty) to a specific product.

Information on the technology: STW artikel ‘Verborgen hygiëne- en kwaliteitsdefecten detecteren met spectrale camera’s’ (zie ookwww.flandersfood.com/stwnieuwsbrief). 

Partners: Imec-NVISION group and MeBioS-KU Leuven

Why participate?: Feasibility tests for thorough quality inspection, visualization of product composition, as well as detection of foreign objects in raw materials, intermediate and final products and/or processes are possible. These tests will be followed by further optimization and a tailor-made suggestion with considerations on the balance between cost and detection performance. After validation, extensional advise for implementation is offered

How to participate? You have to be or become member of Flanders’ FOOD and contribute  (together with the other participating companies) 20% of the project budget. The validation project is also open for SME’s. 

More information on this validation project, the work plan and the financial contribution can be found here