Image quality parameters for user support in industrial computed tomography

• Applicant: Professorin Dr.-Ing. Gisela Lanza
• Subject Area: Production Systems, Operations Management, Quality Management and Factory Planning
• Term: since 2021
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 465978608

The present project deals with industrial X-ray computed tomography (CT). This non-destructive method is particularly versatile in regards to the measurement task. This is due to information being available on both the component surface and the distribution of the density inside a specimen. That this information is freely available in three-dimensional space is particularly valuable in comparison to conventional measurement technology, such as tactile coordinate metrology. In recent years, there has been an increase in interest in industrial CT technology. The large number of use cases presented in literature underlines this. Literature addresses a need to reduce the complexity of CT technology, though. To cover this concern, user support systems are developed. They assist a user to choose optimal settings. In this project, user support systems are enhanced. For this purpose, a method to assess settings is developed. The method uses the image quality of reconstructions to evaluate the goodness of chosen settings. This method may be used in user support systems to evaluate the settings which were recommended by the user support system. Based on this evaluation of the setting recommendation, a user support system can learn. For this purpose, information on a specimen, the setting recommendations, and the associated goodness of the setting are stored as cases. For future setting recommendations, only cases with good ratings are used to recommend new settings. Test specimens are developed and manufactured in various sizes and materials to determine a model to assess settings. These test specimens combine the advantages of conventional test specimens with the radiographic properties of industrial work pieces. The test specimens are used to carry out experiments with varied settings. The variation of the settings follows a method that ensures full coverage of the measurement range of a CT device. Based on the experiments, the image quality-based model for assessing settings is derived. Regression techniques and data-driven methods are deployed to determine a relationship between image quality and the goodness of the settings. The goodness of the settings is represented by the measurement uncertainty and metrological structural resolution. Lastly, the setting assessment model is used in a user support system based on the Case-Based Reasoning methodology. The learning capability is tested by measuring several industrial components multiple times with the user support system to evaluate if repetitions improve the results.

• DFG Programme: Research Grants