TWI Industrial Member Report Summary 1001/2011
By M Amos
Background
In X-ray CT an object is placed between an X-ray source and a digital radiation detector. The object is rotated so that X-ray projection images are acquired from all angles. A filtered back projection (FBP) algorithm is applied to the projection data (also known as an image stack) to compute the attenuation coefficients of the material either as a 3D volume, or a cross sectional slice. The attenuation coefficients relate directly to the material density.
For conventional CT there are several requirements for the acquisition of projection data to ensure accurate reconstruction of the object's attenuation coefficients. One requirement is that the object must, for all projections, fit within the detector's field of view (FOV). In region of interest (ROI) CT the object does not fit within the FOV and the projection data is truncated. Direct reconstruction of this projection data using the common FBP type algorithms will result in imaging errors caused by the missing data and truncation artefacts. These combined errors can mask significant information about the object and can stop accurate density measurements or volume segregation taking place.
Data completion methods can be applied to the projection data so the ROI can be effectively imaged without significant degradation of the CT image quality. The basic concept of a data completion method is to approximate the missing portion of the projection data prior to reconstruction. This paper focuses on the development of two such data completion methods: A cosine extension method and an estimation from model method.
ROI CT can be used either to image objects that are too large to fit within the maximum FOV for a system, or to apply high geometric magnification to a specific area of an object. This investigation has focused on the development of ROI CT for inspection of carbon-fibre-reinforced-plastic (CFRP) panels. CFRP panels benefit from this technique in particular because they can be very large, but need high magnification to image defects such as fibre breakage and matrix cracking.
A 50x50x5mm CFRP panel was fabricated with five holes laser drilled into the surface to act as an image quality indicator (IQI). The holes had sizes of 500, 250, 100, 50, and 30?m.
Objectives
- Determine the source of errors in ROI reconstructions.
- Develop a data completion method for fan beam geometry that overcomes the truncation artefacts associated with ROI CT inspection of CFRP laminates.
- Develop a data completion method for fan beam geometry that overcomes all reconstruction errors associated with ROI CT inspection of CFRP laminates.
- Quantitatively validate the performance of both data completion methods.
- Implement the data completion methods for cone beam ROI CT.