Automated machine vision system for film analysis and defect recognition

Development of an imaging system to use for conductive and decorative film defect recognition. The system, placed in the production environment, will be used to analyze printing patterns, as well as to recognize possible defects happening in printing. The system is capable of recognizing between different defect types and quantifying their severity. Ultimately it should be able to provide also feedback to the operator with device tuning parameter values.

Tool is divided into modules:

• Camera image pre-processing
• Image segmentation
• Feature recognition and computation of feature vectors
• Segment classification

The tool is capable of recognizing 5 typical defects in conductive ink screen printing:

• broken trace,
• bumps,
• hair,
• saw pattern,
• thinning of a trace.

Solution is based on python, therefore python runtime with required libraries is required. For image acquisition, high quality camera system with controlled lighting is required.

End users of the Tactotek's manufacturing process. This system ultimately will become a software-hardware package offered as a service.

Not yet concrete plans to establish conditions of reuse, because the tool is still under development.