Use this URL to cite or link to this record in EThOS: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.656338
Title: Analysis of tomographic images : quantitative methods for soil pore structure
Author: Houston, Alasdair Neil
Awarding Body: University of Abertay Dundee
Current Institution: Abertay University
Date of Award: 2013
Availability of Full Text:
Access through EThOS:
Abstract:
Soil is a material of tremendous importance to the sustainability of complex life on planet Earth and X-ray computed micro-tomography permits the minimally invasive study of 3D soil pore structure at appropriate scales. However, the tremendous variety of materials and their complex spatial arrangement that may be found within soil, in addition to the vast quantity of information present within a tomographic image, lead to significant difficulty in the accurate interpretation of these images. This thesis deals with problems of acquisition, assessment, processing and measurement for such images, applying unsupervised (fully automatic) methods as a means of dealing with very large data in a consistent fashion. Existing methods of image analysis are assessed, refined, extended and applied in novel ways to meet the challenges posed by soil, Noise resulting from X-ray scattering is a significant problem when samples having a substantial mineral content (i.e. most soils) are the subject of tomographic imaging. Further compounding this problem is the high degree of heterogeneity seen in the spatial distribution of material within soil, covering scales from the sub-micron level up to those obvious to the naked eye. Reliable automatic identification of pore versus solid structure requires sensitivity to coherent evidence yet robustness against noise. It is shown that this very significant technical challenge is robustly addressed via the consideration of spatial correlation, but this involves potentially very high computational cost . This latter problem is addressed through developing a novel algorithm that extends the indicator kriging approach, greatly reducing computational cost and simultaneously improving the quality of results.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.656338  DOI: Not available
Share: