Discriminate between seeds of cereals using image analysis (BAGRAANZAAD)

Using image analysis, is it possible to discriminate between seeds of different cereals and to check grain seed lots to see if they match the certification standard of only 1% other species? The purpose of this study is to use the image analysis technique to classify seeds of different cereals and determine how feasible, efficient and correct the classification is.  It is legally required to test seed lots for contamination with other species or impurities. Traditionally laborious work has been done by experts, but often closely related species are hard to distinguish.

For this study different varieties of oats, winter barley, rye, winter wheat, spelt and triticale were used. For each variety, 1000 seeds are judged based on form and color using image analysis. By means of discriminant analysis, parameters or combination of parameters were found that allow a good classification of the different species.

By making use of the fast-developing image processing possibilities, we intend to reduce the dependence on visual assessment by experts. Difficult and time consuming activities (eg. control of seed lots) can be performed in a more efficient and labour-friendly way.