High-tech and Artificial Intelligence Make the Invisible Visible

Research and Development in Biosystems Engineering

Quality Assurance in Food Production

Non-invasive identification of coffee by means of hyperspectral measurement is one example of a number of projects on quality assurance in food production. The Biosystems Engineering Expert Group is working on other contract research projects for a variety of clients.

Quality conscious producers make great efforts to assure quality when roasting high-grade coffee. Noninvasive methods are needed to inspect the quality of raw coffee being processed continuously and to monitor roasted coffee continuously. Methods based on optical measurement provide considerably more comprehensive quality assurance than conventional sampling. If required, the methods of measurement and analysis developed by the Biosystems Engineering Expert Group can be integrated directly in the production process. Hyperspectral measurements and the analysis of thusly obtained extremely high dimensional data by methods of machine learning make it possible to analyze any relevant constituents.

When relevant reference data is available, coffee species, blend ratios and constituents can be quantified. The system is neither designed nor able to attain the precision of specialized laboratory tests. Rather, it is a cost effective, real-time capable alternative rapid test.

Retailers and consumers can also use the system to inspect promised quality attributes, e.g. coffee species, special roasting, untreated condition, etc., without complex and costly laboratory tests.

Retailers and consumers can also use the system to inspect promised quality attributes, e.g. coffee species, special roasting, untreated condition, etc., without complex and costly laboratory tests.

The Department of Analytical and Organic Chemistry at Jacobs University Bremen and the Department of Mathematics at Mittweida University of Applied Sciences are collaborating with us on this research.

© Fraunhofer IFF

Laboratory setup for the measurement of relevant constituents in coffee samples with a hyperspectral camera using artificial broadband illumination and a motorized translation unit.

Publications

 

 

Backhaus, A.; Bollenbeck, F.; Seiffert, U.: High-Throughput Quality Control of Coffee Varieties and Blends by Artificial Neural Networks from Hyperspectral Imaging. Proceedings of the First International Congress on Cocoa, Coffee, and Tea CoCoTea 2011, 88-92

Backhaus, A.; Lachmair, J.; Rückert, U.; Seiffert, U.: Hardware Accelerated Real Time Classification of Hyperspectral Imaging Data for Coffee Sorting. Proceedings of the 20. European Symposium on Artificial Neural Networks ESANN 2012, 627-632