Erosion and dilation of edges in dimensional X-ray computed tomography images

Dimensional X-ray Computed Tomography (CT) is a rapidly expanding field of research due to the numerous advantages this technique offers over conventional measurement technologies, most notably, the ability to measure internal features of a component. Tactile and optical Coordinate Measurement Machines (CMM), currently used in the manufacturing production industry, record points on the external surface of a workpiece by measuring the contact point of a physical probe or the reflection of projected light. X-ray CT has the ability to capture full volumetric data, since X-rays are transmitted through the entire object, revealing features which are otherwise invisible. Over the past five years, interest in this field has grown in the UK, with an increasing number of organisations in industry and research having access to X-ray CT machines and the wide range of manufacturers, offering new systems specifically designed for dimensional metrology applications.

Despite this, the complexity of data acquisition required for dimensional measurement using X-ray CT has made it difficult to estimate the measurement uncertainty. This has hindered the generation of standards and full-scale adoption of this technique in industry. Due to the nature of X-ray imaging, a number of non-linear influence factors exist which have the potential to cause dimensional measurement error. These influences must be better understood to reduce and ideally, compensate error.

In this doctoral thesis, the effects of the influence factors associated with CT data acquisition are studied, specifically, beam hardening and a finite X-ray source size. The effects these have on the quality of X-ray CT data are well understood; typically degrading the achievable contrast and spatial resolution of the CT image. However, the effects on dimensional measurement are less well understood due to the complexity of their interactions before reconstruction of the final image. These influences are modelled in a simulated CT acquisition to quantify any systematic effects on determination of edges in the CT image. The results are then validated by experimentally replicating the simulation set-up.

In this work, it is found that beam hardening and a finite source diameter can lead to systematic errors in the edge position within the CT image. Beam hardening generally leads to dilation of the edge; where the edge position moves in the direction of the surface vector. In contrast, a finite source diameter can lead to erosion of the edge; where the edge position moves in an opposing direction to the surface vector.