The challenge

A company in the printing industry, producing packaging for the pharmaceutical industry, needs to automate quality processes.

The company has a very high volume of rejects due to the quality of the finished product because of incorrect printing. Mostly in case marking defects related to text and braille.

These errors can appear in the braille die-cutting, in the printing with ink, or due to errors in the PDFs that will be used to create the plates for printing the cases.

Objectives

Implement a machine vision system within the quality system.

The machine vision environment consists of a scanner for image acquisition, and computers for image processing and comparison with the models.

The scope of the project is limited to the development of the software tool for image acquisition, processing and comparison, excluding the equipment supporting the operation of the tool.

How have we done it?

With a scanning + docker application. The tool is installed on Windows computers, requiring the Docker tool to be able to run it. It is configured to save the processed image files and reports in the cloud environment of the graphics industry.

And all this thanks to the technologies:

• Template matching
• Layout comparison
• Local and global image alignment
• Deep learning

Results provided

The developed solution facilitates the early detection of these errors and reduces waste, as well as increasing customer confidence in the developed product.