Use this URL to cite or link to this record in EThOS: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.700449
Title: Medialness-based shape invariant feature transformation
Author: Aparajeya, Prashant
ISNI:       0000 0004 5993 481X
Awarding Body: Goldsmiths, University of London
Current Institution: Goldsmiths College (University of London)
Date of Award: 2016
Availability of Full Text:
Access through EThOS:
Full text unavailable from EThOS. Thesis embargoed until 30 Sep 2019
Access through Institution:
Abstract:
This research is about the perception-based medial point description of a natural form (2D static or in movement) as a generic framework for a part-based shape representation, which can then be efficiently used in biological species identification, as well as more general pattern matching and shape movement tasks. We consider recent studies and results in cognitive science that point in similar directions in emphasizing the likely importance of medialness as a core feature used by humans in perceiving shapes in static or dynamic situations. This leads us to define an algorithmic chain composed of the following main steps. The first step is one of fuzzy medialness measurements of 2D segmented objects from intensity images that emphasizes main shape information characteristic of an object's parts, e.g. concavities and folds along a contour. We distinguish interior from exterior shape description. Interior medialness is used to characterise deformations from straightness, corners and necks, while exterior medialness identifies the main concavities and inlands which are useful to verify parts extent and reason about articulation and movement. The second main step consists on defining a feature descriptor, we call ShIFT: Shape Invariant Feature Transform constructed from our proposed medialness­based discrete set, which permits efficient matching tasks when treating very large databases of images containing various types of 2D objects. Our defined shape descriptor ShIFT basically captures elementary shape cues and hence it is able to characterise any 2D shape. In summary, our shape descriptor is strongly footed in results from cognitive psychology while the algorithmic part is influenced by techniques from more traditional computer vision.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.700449  DOI: Not available
Share: