Quality inspection software

Quality inspection is typically a very precise and time-consuming stage within the production process. At the same time, this step is of utmost importance. Falsely approved items do not meet the set tolerances, which could cause a misfit for its intended use, or worse – such as safety issues. To prevent human errors, to meet a high level of accuracy, and to speed up the process, it is vital to have this production stage automated.

To automate their quality inspection process, Mintres BV reached out to the FIP-AM@UT. Mintres BV is a company that specialises in fabricating components in advanced ceramic materials such as diamond and aluminium nitride, mainly for use in the electronics and opto-electronics industry. To streamline their production process, they were looking for an automation of their quality inspection and control process.

The goal of the project was to fully automate the quality inspection with an advanced vision-assisted setup, by using image analysis algorithms for detecting defect classes. With this solution, accuracy of the inspection process is increased.

Industry partner


To develop a fitting software solution for Mintres, the project was divided in three phases with specific sub goals and deliverables. The results of each phase inform the following steps of the project to create a successful final result.

Phase 1: Research of use case for understanding of microscopic defect and production environment.
Phase 2: Development of hardware setup, data acquisition, and evaluation.
Phase 3: Computer vision software development for automated identification of defects.


The collaboration between Mintres and FIP-AM@UT led to the development of a computer vision algorithm software designed to meet Mintres’ quality assurance needs. The project resulted in:

  • A report on selected products, defect classes and features definition
  • Operating modular software libraries that automatically identify surface defects on the chosen product
  • Operating GUI which allows to choose from desired parameters for the inspection algorithm such as product, defect type to be identified, and quality tolerance
  • Transfer of Knowledge to the employees through a training session

This software can identify up to six defects (both dimensional and surface defects), offers a detection resolution in the order of a few microns and robust detection greater than 10 microns, and can scan 11 samples per minute (including all 6 defect classes).

“It has been a fruitful collaboration. FIP-AM@UT has significant knowledge in this area and has proven to be a flexible partner and has handled all the project planning changes required when working in a highly dynamic and fast-moving market.”

Clive Hall
Co-founder & Managing Director Mintres B.V.

For more information, feel free to reach out to:

Hari Subramani

Hari Subramani Palanisamy

Research Engineer