Automated defect detection for fast evaluation of real inline CT scans

ABSTRACT

In this paper, we present a fully automatic evaluation approach that can be used for fast inline CT scanning. In contrast to classical defect-recognition algorithms, this approach does not require good image quality. Instead, it allows to distinguish CT artifacts from real defects introduced in the production process. To this end, a three-step workflow was developed, in which any deviations from a reference part are detected and subsequently classified and segmented, to allow automatic decision whether the part fulfills given quality requirements. We demonstrate this approach using CT scans of aluminum castings, typically used in the automotive industry. Also, a comparison between two different segmentation algorithms is shown.

KEYWORDS:

• Inline computed tomography
• automated defect detection
• machine learning

Acknowledgements

Part of this research has been funded by the German Federal Ministry of Education and Research project BMBF 02K16C082 Produktionsbezogene Dienstleistungssysteme auf Basis von Big-Data-Analysen (ProData). We thank our colleagues from ZEISS Industrial Quality Solutions for helpful discussions and for proving insights and expertise. We thank our Master’s student Simon Krafft for assistance in preparing figures.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by the Bundesministerium für Bildung und Forschung.