Abstract
In the field of medicine, X-ray Computerised Tomography (CT) has been applied to determine the density variation within a body non-destructively. The internal structure of many industrial objects is also of great importance, for example the wing of an aircraft. Thus CT is currently used in an industrial environment. This thesis investigates the methods by which the scanner architecture and image reconstruction software can be optimised for real-time computerised tomography of industrial objects using parallel computing. Since industrial objects are generally dense the choice of radiation source used to generate projection profiles of such objects was examined. The speed of image reconstruction is largely dependant on two factors; one is the speed at which data can be acquired from the scanning system and the other is the speed at which such data may be reconstructed to yield an image. The speed of data acquisition can be enhanced by using a configuration of multiple detectors. This requirement has a great bearing on the choice of architecture of the scanning system. Image reconstruction is a computationally intensive task, it therefore requires novel processing techniques to maximise the efficiency of execution. This can be accommodated by implementing the reconstruction algorithm on a multiple processor system. A scanning system that contains multiple detectors for data acquisition and more than one processor to implement image reconstruction is complex, the method of control of such hardware is described within this work. The images reconstructed from data acquired using a purpose built fan beam scanner are the subject of qualitative and quantitative analysis. The quality of the images is related to the scan parameters architecture of the fan beam scanner. A discussion of future work is included which outlines ways in which the scanner hardware and software may be improved to further enhance the speed of real-time image reconstruction.