Machine Vision Combined with Near-Infrared Spectroscopy to Guarantee Food Safety A high-speed, single-kernel robotic instrument has been developed to analyze the composition and physical characteristics of individual kernels of grain and sort them according to specified parameters. Although the automated analysis has been applied to several different types of grains, seeds, and beans, this article focuses on gluten-free ready-to-eat breakfast cereals. Oats are a gluten-free grain but can be contaminated with wheat, rye, or barley during farming, storage, transport, or other stages in the supply chain. To test for contamination, food processors must either manually check grain samples or analyze them using wet chemistry methods. These tests are both costly and time-consuming. Attempts have been made to use machine vision in analyses in order to reduce the cost and time requirements for these analyses. These attempts have failed, however, due to the similarities between various cereal grains with regard to the parameters tested—similarities that also make it difficult for human inspectors to accurately assess samples. The new high-speed, single-kernel instrument makes it possible to analyze kernels using both color image and spectral analyses. With the addition of spectral analysis, 40 kernels/sec can be classified, with higher than 95% accuracy and very low repeatability errors. |