Bringing the Computed Tomography (CT) technology into production lines brings many advantages due to the complete and detailed 3D characterization and localization of defects. But compared to laboratory systems, Inline-CT systems must meet several challenges, the most significant one is certainly the limited available time for the measurement procedure and data evaluation. Visual inspection of 3D data sets of complex parts within the production rate is almost impossible, so automated image evaluation algorithms are mandatory for that task. Fast measurement procedures often mean short integration times and a lower number of projections which are two factors that increase quantum noise and thus decrease image quality significantly. In addition to that, scatter artifacts also have a negative influence on the image quality. All these conditions constitute challenges to the image evaluation algorithm. The Fraunhofer EZRT has developed a method for automated defect detection in cast parts by using reference (that means flawless) parts for comparison. The advantage of this method is that image artifacts are considered, since they appear in both data sets in the same way and have almost no negative influence on the reliability of the evaluation result. Most recent work took this method one step further by using not real flawless parts, but instead using simulations to generate reference data sets in a very convenient and quick way for any number of different parts.
The exact knowledge of the geometric acquisition parameters directly influences the quality of a tomographic reconstruction. Most calibration procedures use calibration phantoms that consist of metal beads arranged in a specific order. If the exact positions of the metal beads are to be sufficiently known, the production of such calibration phantoms can become very expensive. To avoid this problem, we propose a method that uses single plastic bricks (e.g. LEGO bricks) as calibration phantom. Based on a CAD model, projections with given acquisition parameters are simulated and then compared to the real projection using a metric. The objective of our research is to evaluate which metrics are best suited for the comparison of such projections. The results of applying two metrics on simulated and real data are presented.
Today, Computed Tomography is becoming increasingly important as a non-destructive testing technique in industry. The advantages of evaluating 3D information are manifold and system costs have fallen in recent years. However, these systems suffer from the disadvantage that performing a measurement is more complex and error-prone than pure radioscopic testing. Thus, users with more skill and experience are needed to get the most out of the scan. Nowadays, the process of parametrizing the CT scan is often completely manual and mostly performed by inspection engineers due to the high number of available parameters. These are spatial resolution, magnification, object and detector distances, number of projections and, most importantly, positioning of the specimen on the turntable. Choosing appropriate parameters is not trivial due to the interplay between required image quality and scan time. So, identifying optimal parameters can be a major challenge. The main goal of the scan planning is to solve the scan task dependably, process-safely if applicable, and efficiently with minimal investment of time. Simulation-based determination of the optimal parameters can enable objective parameters to be obtained that are less depended on expert users. This is especially important when a wide variety of objects need to be scanned. 3D printed parts, which can have a batch size of one under extreme conditions, serve as an example. The range of parts produced is extremely wide and complex, which leads to a complex configuration of the scan. Fraunhofer EZRT is developing software solutions to find an ideal scanning trajectory for a given scan task. For this purpose, CAD data of the part is used to simulate CT reconstructions and to evaluate the suitability of the parameters in focus of the task. For this, the image quality of the component, for metrological task e.g., and the detectability of faults inside the material, pores e.g., are considered. The user is given the ability to provide the parameter space especially with the regard to the geometry of the system. Also, the user can define quality criteria, pore sizes in different component locations e.g., which are considered in the optimization process. In the first step, those parameter sets are selected which have the best ratings on a projection basis. Subsequently, reconstructions are made on the remaining candidates, and these are evaluated. This approach saves computational time, the entire parameter space can be sampled, and a meaningful evaluation of the expected reconstruction can be performed.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.