Orca MoldControl was originally developed to monitor injection moulds with regard to their correct and operational status.

This could mean either the correct positioning of inserts or the presence of foreign bodies inside the cavity.

The measurement method that is used here explicitly extracts three different features from image contents, and compares each one separately in a higher-dimensional space. This involves calculating features to describe the form, structure and optical characteristics.

The first stage involves using these features for the online training of a one-class classifier, which also allows for the training of varying setpoint statuses. A second training stage offers the option of training a binary network classifier through the presentation of negative examples. This ultimately makes it possible to reliably detect even the slightest differences.

The separation of the features means that their influence on the overall result is infinitely variable, making it possible to achieve invariance to certain perturbations.

As a result, Orca MoldControl is an easy-to-integrate system for the automatic monitoring of production sequences. The continual further development of the system means that it is also becoming increasingly suitable for fully automatic quality assurance.