Digital Imaging Based Classification and Authentication of Granular Food Products

In the food industry there are many types of product that are in the form of particles, granules or grains. Consistent material size and quality within any given sample is an important requirement that is well known in industry. In addition it is possible that samples of material may be of unknown type or have been subject to adulteration, thus making material authentication a real requirement. The present work implements an advanced, but cost-effective, digital imaging and image processing technique to characterize granular foodstuffs either in real time process control or in an off-line, sample-based, manner. The imaging approach not only provides cost-effective and rugged hardware when compared with other approaches but also allows precise characterization of individual grains of material. In this paper the imaging system is briefly described and the parameters it measures are discussed. Both cluster and discriminant analyses are performed to establish the suitability of the measured parameters for authenticity study and a simple fuzzy logic is implemented based on the findings. Tests are performed, using rice as an example, to evaluate the performance of the system for authenticity testing, and encouraging results are achieved