Use this URL to cite or link to this record in EThOS: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.638674
Title: Video genre classification
Author: Roach, M. J.
Awarding Body: University of Wales Swansea
Current Institution: Swansea University
Date of Award: 2003
Availability of Full Text:
Access from EThOS:
Abstract:
The thesis progresses to look at feature extraction and description for content based analysis involving audio and visual modes. The notion calculation is itself very simple - just a summation of the difference at points in two images. The measure is filtered using a simple digital filter whose output is analysed by Fourier. This is then processed to derive feature vectors from spectral and from regressive analysis. I do have some reservations here - it would appear that the classified data are aliased and there is no discussion of appropriate strategies to mitigate this, or as to whether it leads to untoward effects. I do note that one genre appears largely to be noise. I do have reservations beyond technical ones for it does appear that some of these points could be discussed within the papers bound into the thesis, but I seek clarification and guidance on these points at viva voce. I shall also seek information concerning how the parameters were set since this would appear important, yet some aspects in this choice remain unclear. The research does appear to have developed one of the more demanding data sets for experimental evaluation. This claim could have been amplified, or even made. That new technique appears to have been deployed to success would appear to argue against my technical concerns, but there is an issue of generalisation capability, or the new approach's likely capability of categorising correctly data that is as yet unseen but has been presented in an appropriate format. I certainly consider that the results should be strengthened by an analysis of statistical significance. I certainly agree that there is considerable scope for future work in this area. Mr Roach is also to be congratulated on a good record on conference publication but I do have some reservations of a technical nature for which I shall seek clarification at viva voce.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.638674  DOI: Not available
Share: