Sorting equipment of the type OptoSort® is learning. With the help of the system function „training “by the user after identical color characteristics selected objects are driven over the sorting equipment and registered as color class. In addition it is enough to drive some good material in the simplest case first somewhat over the sorting equipment, followed by a small quantity of bad material.
Already finest differences in color can be recognized by the sensitive camera system and the training procedure. Up to eight different and freely parametrable color classes can be trained and weighted. By the help of a probability distribution by Bayes outliers in the trained material are eleminated in the assigned decision table. 

By morphologic functions the material admission can be improved besides, disturbing particles and noise for example will removed or the weighting of hotspots strengthened.

The sort decision whether an object is good or bad is meet by the system due processing a programable decision chain. Decisions are possible on the basis of the color surfaces, color composition and geometrical characteristics. Surface thresholds for the color classes are typical, whereby each class can still be linked logically if necessary for recognize complex color compositions, for example with quartz.



Depending upon installation type only a few milliseconds pass between the detection of an object by the camera system and the entrance into the range of the ejector bar, this offense ensure a precise removing with little product lost. As a result of a material-specific selection of the valves, inserted with small distance to the flight path of the material and the accurate object recognition and control the compressed air consuption is a very small.