ePrints.FRI - University of Ljubljana, Faculty of Computer and Information Science

Multiresolution parametrization for texture classification and its use in the scintigraphic image analysis

Luka Šajn (2007) Multiresolution parametrization for texture classification and its use in the scintigraphic image analysis. PhD thesis.

[img] PDF (Disertacija)
Download (1320Kb)


    In the dissertation multiresolutional texture parametrization is addressed and the original algorithm ARes for finding more informative resolutions in the sense of classification accuracy is proposed. Our study explores the multiresolutional texture parametrization approach based on the image content with regard to the parametrization quality, especially in case of the ArTex algorithm. The tested parametrization algorithms using multiresolutional approach have demonstrated significant improvements in results over one scale parametrization. This supports the hypothesis that the resolution selection is important for texture parametrization. The developed algorithm ARes in combination with the ArTex algorithm achieves statistically significant improvements over single resolution and also over equidistant resolutions. We have confirmed that the use of the equidistant resolution space when parameterizing textures significantly outperforms the use of the exponential resolution space, which is used by majority of authors. For the multiresolution parametrization applicative domain two medical cases have been used, sequential diagnostics of coronary artery disease and diagnostics of whole-body bone scintigraphy. The whole-body scintigraphy segmentation process is presented. The presented computer-aided system for bone scintigraphy is a step towards automating the routine medical procedures. In the case of coronary artery disease we have shown that multi-resolution ArTeX parametrization using machine learning techniques can be successfully used as an intelligent tool for image evaluation, as well as as a part of the sequential diagnostic process which is considerably improved.

    Item Type: Thesis (PhD thesis)
    Language of Content: Slovenian
    Mentor / Comentors:
    Name and SurnameIDFunction
    prof. dr. Igor Kononenko237Mentor
    Link to COBISS: http://www.cobiss.si/scripts/cobiss?command=search&base=50070&select=(ID=6128468)
    Institution: University of Ljubljana
    Department: Faculty of Computer and Information Science
    Item ID: 707
    Date Deposited: 08 Dec 2008 18:55
    Last Modified: 13 Aug 2011 00:34
    URI: http://eprints.fri.uni-lj.si/id/eprint/707

    Actions (login required)

    View Item