Industrial Computed Tomography (CT) is one of the latest technology in the field of dimensional metrology and in the last decade has been rapidly growing due to its versatility. Dimensional measurements with CT are not limited only to the workpiece's surface as the conventional measurement methods. Indeed, it is possible to evaluate volumetric models and therefore measure the inner part of a workpiece without destroying it. Several CT parameters can influence the overall quality of the reconstructed volume and thus measurement results. These quantities, defining the goodness of the measure, are currently subject of research. From an industrial point of view finding the optimal CT setting is crucial to be competitive. Among all parameters, those that have a direct impact on scanning time are the most important ones. In this work the most influencing variables are divided in two groups: those ones that do not have influence on scanning time and those ones that have. The time-independent variables, like the orientation of the workpiece in the CT system, the Source-Object distance, the voltage and the current play an important role and they must be optimized. CT users choose values for these parameters according to their experience, since the complexity of CT and the lack of internationally recognized standards and guidelines make the choice of the acquisition parameters a time consuming iterative process. This often leads to high variability in measurement results, insufficient accuracy or unnecessarily long scanning time. To solve this problem, we present and validate a model able to optimize time-independent parameters. The model is based on a ray-tracing algorithm. Once optimized the time independent variables a method to reduce scanning time, minimizing the number of projection is presented and validated. The number of projections is an important factor to define the ability of the CT system to spatially resolve the geometrical features of a workpiece and it mainly defines CT scanning time. Low numbers of projections cause artifacts due to under-sampling to occur, whereas acquiring too many projections involves high scanning times. In conclusion, the focus of this model is on minimizing the scanning time while keeping the best possible measurement accuracy.
La qualità complessiva del volume ricostruito, e quindi le misure ad esso associate, dipendono da diversi parametri. La definizione e l'ottimizazione di questi parametri sono l'oggetto di questa ricerca. Alcuni di questi parametri hanno un impatto diretto sul tempo di scansione e quindi sul costo e versatilità della CT in contesti industriali. Nello sviluppo di questo lavoro verranno divise le variabili in due gruppi: tempo-dipendenti e tempo-indipendenti. Tra i parametri tempo indipendenti, l'orientamento del pezzo nel sistema CT, la distanza tra sorgente-oggetto, la tensione e la corrente giocano un ruolo importante e necessitano di essere ottimizzati. Allo stato dell'arte gli operatori CT scelgono i parametri in base alla loro esperienza, poiché la complessità della CT e la mancanza di norme e guide internazionalmente riconosciute rendono la scelta dei parametri di acquisizione un processo lento e iterativo. Questo porta spesso a elevata variabilità dei risultati di misura, accuratezza insufficiente o tempi di scansione improduttivi. Per risolvere questo problema, viene presentato e validato un modello in grado di ottimizzare i parametri tempo-indipendenti. Tale modello si basa su un algoritmo di Ray-tracing. Una volta ottimizzate le variabili tempo-indipendenti, verrà presentato e validato un modello per ridurre il tempo di scansione, minimizzando il numero di proiezioni senza sacrificare la qualità della scansione stessa. Il numero di proiezioni é infatti un fattore importante che definisce la capacità di un sistema CT di "risolvere" spazialmente le caratteristiche geometriche di un pezzo. Un basso numero di proiezioni causa errori dovuti al fenomeno del sotto-campionamento, mentre l'acquisizione di un numero eccessivo di proiezioni comporta tempi di scansione inutilmente elevati. L'obiettivo é quindi quello di ridurre al minimo il tempo di scansione mantenendo la migliore accuratezza di misura possibile.
Feature based method to optimize computed tomography settings for fast and accurate analysis
BEN ACHOUR, SOUFIAN
2015/2016
Abstract
Industrial Computed Tomography (CT) is one of the latest technology in the field of dimensional metrology and in the last decade has been rapidly growing due to its versatility. Dimensional measurements with CT are not limited only to the workpiece's surface as the conventional measurement methods. Indeed, it is possible to evaluate volumetric models and therefore measure the inner part of a workpiece without destroying it. Several CT parameters can influence the overall quality of the reconstructed volume and thus measurement results. These quantities, defining the goodness of the measure, are currently subject of research. From an industrial point of view finding the optimal CT setting is crucial to be competitive. Among all parameters, those that have a direct impact on scanning time are the most important ones. In this work the most influencing variables are divided in two groups: those ones that do not have influence on scanning time and those ones that have. The time-independent variables, like the orientation of the workpiece in the CT system, the Source-Object distance, the voltage and the current play an important role and they must be optimized. CT users choose values for these parameters according to their experience, since the complexity of CT and the lack of internationally recognized standards and guidelines make the choice of the acquisition parameters a time consuming iterative process. This often leads to high variability in measurement results, insufficient accuracy or unnecessarily long scanning time. To solve this problem, we present and validate a model able to optimize time-independent parameters. The model is based on a ray-tracing algorithm. Once optimized the time independent variables a method to reduce scanning time, minimizing the number of projection is presented and validated. The number of projections is an important factor to define the ability of the CT system to spatially resolve the geometrical features of a workpiece and it mainly defines CT scanning time. Low numbers of projections cause artifacts due to under-sampling to occur, whereas acquiring too many projections involves high scanning times. In conclusion, the focus of this model is on minimizing the scanning time while keeping the best possible measurement accuracy.| File | Dimensione | Formato | |
|---|---|---|---|
|
2016_04_BenAchour.pdf
non accessibile
Descrizione: Feature based method to optimize Computed Tomography settings for fast and accurate analysis
Dimensione
13.79 MB
Formato
Adobe PDF
|
13.79 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/121529