Pulsed terahertz (THz) systems are an emergent technology, finding diverse applications as they approach maturity. From their birth in the late 1980's to the wealth of alternate sources and imaging modalities now available, the rise has been fuelled by the expectation that this will prove a world changing technology. This Thesis takes an application focused approach and seeks to provide enabling systems and algorithms for the development of functional imaging systems with broad potential application in security inspection, non-destructive testing and biomedical imaging. Three dimensional pulsed THz imaging systems were first introduced in 1996 using a reflection-mode ultrasound-like configuration. This Thesis builds upon this former work by focusing on transmission mode tomography systems using pulsed THz radiation. Several novel 3D imaging modalities are introduced. The hardware architectures, based on optoelectronic generation and detection of THz radiation are described. Approximations to the wave equation are derived, allowing linear reconstruction algorithms to recover 3D structural information fromthe transmitted THz field. Finally the systems are demonstrated and the achievable resolution and image quality are investigated. Three imaging architectures are developed herein: 1. T-ray holography allows the 3D distribution of point scatters to be resolved based on a single projection image utilising a novel reconstruction algorithm based on the windowed Fourier transform and back-propagation of the Fresnel-Kirchhoff diffraction equation. 2. T-ray diffraction tomography utilises the diffracted THz field to allow a Helmholtz equation based, frequency-dependent reconstruction to be performed and the THz spectrum at each pixel to be calculated. 3. T-ray Computed Tomography (CT) uses analogous techniques to X-ray CT, based on the Radon transform, to provide 3D T-ray reconstructions of unprecedented fidelity. These techniques have important applications in material identification, which is investigated in the second part of this Thesis. Pulsed THz spectroscopy has been widely acclaimed for its potential to identify different materials based on their spectral properties. The second part of this Thesis presents algorithms towards this goal. Three case studies are performed focusing on biomaterial classification, anthrax detection and in vitro osteosarcoma cell differentiation. A classification framework is developed to process the THz spectral data and identify specific materials. A linear filter model is introduced to describe the system response of different materials, and the filter taps are utilised for feature extraction. This technique is demonstrated for biomaterial and anthrax classification. For cell differentiation a genetic algorithm is used to select deconvolved frequency components to train a classifier. In each case a high classification accuracy is demonstrated, highlighting the promise and potential of three dimensional T-ray inspection systems. / Thesis (Ph.D.)--School of Electrical and Electronic Engineering, 2004.
| Identifer | oai:union.ndltd.org:ADTP/263851 |
| Date | January 2004 |
| Creators | Ferguson, Bradley S. |
| Source Sets | Australiasian Digital Theses Program |
| Language | en_US |
| Detected Language | English |
Page generated in 0.0024 seconds