Abstract
dc:description.abstract<p>Active contour models, or snakes, have become a well established technique in the image processing community. This field has been a fertile area of research, producing many and varied extensions to the original Kass snake. Of primary importance to this thesis are the extensions which incorporate motion. This work presents a novel approach to incorporate velocity into the formulation of an active contour model. Whereas most time based techniques are centred around a tracking framework, we present a new model off'ering a single description of the target under investigation and its velocity rather than a series of snapshots. The active contour models we have developed shows the advantage of incorporating the information across a sequence of frames to extract the boundary of an object moving with constant velocity. The information averaging inherent in the model aids in extracting occluded objects and also performs suitable rejection of static background data. Two models have been developed, firstly a global velocity model to test the efiicacy of the approach and secondly, a local velocity model to address the short-commings of the local velocity model. Keywords: Image Processing, Snake, Velocity, Active Contour</p>
Degree
thesis:*- Name dc:type.qualificationname
- Ph.D.
- Level dc:type.qualificationlevel
- doctoral
- Grantor dc:publisher.institution
- University of Southampton
- Year dc:date.issued
- 2002
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Roddis, Robert
Rights
- Language dc:language
- English, eng